Download Pension Reform and the Development of Pension

Pension Reform and the Development
of Pension Systems: An Evaluation of
World Bank Assistance
Background Paper
Mexico Country Study
Salvador Valdés-Prieto
Director-General, Independent Evaluation Group: Vinod Thomas
Director, Independent Evaluation Group, World Bank: Ajay Chhibber
Senior Manager: Ali Khadr
Task Manager: Emily Andrews
This paper is available upon request from IEG.
2007
The World Bank
Washington, D.C.
ENHANCING DEVELOPMENT EFFECTIVENESS THROUGH EXCELLENCE
AND INDEPENDENCE IN EVALUATION
The Independent Evaluation Group is an independent unit within the World Bank Group; it reports
directly to the Bank’s Board of Executive Directors. IEG assesses what works, and what does not; how a
borrower plans to run and maintain a project; and the lasting contribution of the Bank to a country’s
overall development. The goals of evaluation are to learn from experience, to provide an objective basis
for assessing the results of the Bank’s work, and to provide accountability in the achievement of its
objectives. It also improves Bank work by identifying and disseminating the lessons learned from
experience and by framing recommendations drawn from evaluation findings.
IEG Working Papers are an informal series to disseminate the findings of work in progress to encourage
the exchange of ideas about development effectiveness through evaluation.
The findings, interpretations, and conclusions expressed here are those of the author(s) and do not
necessarily reflect the views of the Board of Executive Directors of the World Bank or the governments
they represent.
The World Bank cannot guarantee the accuracy of the data included in this work. The boundaries, colors,
denominations, and other information shown on any map in this work do not imply on the part of the World
Bank any judgment of the legal status of any territory or the endorsement or acceptance of such boundaries.
Contact:
Knowledge Programs and Evaluation
Capacity Development Group (IEGKE)
e-mail: [email protected]
Telephone: 202-458-4497
Facsimile: 202-522-3125
http:/www.worldbank.org/ieg
Acronyms and Abbreviations
AFORE
CONSAR
CPI
CURP
FOVISSTE
FTAL
GDP
IADB
IEG
IMSS
INFONAVIT
ISSSTE
IVCM
OECD
OED
SAL
SAR
SHCP
Administradora de Fondo para el Retiro (Retirement Fund
Administrator)
Comisión Nacional del SAR (National Commission of the
Retirement Savings System)
Consumer Price Index
Clave Unica de Registro Poblacional
Fondo de la Vivienda del ISSSTE (Housing Fund for the Staff of
the Institute of Social Seurity and Government Workers)
Financial Sector Technical Assistance Loan
Gross domestic product
Inter-American Development Bank
Independent Evaluation Group (formerly OED)
Instituto Mexicano del Seguro Social (Mexican Institute of Social
Security)
Instituto Nacional para el Fomento de la Vivienda para
Trabajadores (National Institute for the Development of Living
Quarters for Workers)
Instituto de Seguridad y Servicios Sociales de los Trabajadores del
Estado (Institute for Security and Social Services for Government
Workers)
Invalidez, Vejez, Cesantia en Edad Avanzada, y Muerte
(Disability, Old Age, and Survivorship)
Organisation for Economic Co-oporation and Development
Operations Evaluation Department (changed its name to IEG in
December 2005)
Structural Adjustment Loan
Sistema de Ahorro para el Retiro (Retirement Savings System)
Secretaría de Hacienda y Crédito Público (Secretariat of Finance
and Public Credit)
Contents
Preface................................................................................................................................. i
1. Background ..................................................................................................................1
Pension System Structure before 1997 ........................................................1
Redistribution toward the Elderly Poor ...........................................2
Savings-Insurance Programs............................................................3
Human Resources Management: Aims and Portability ...................4
The IMSS/ICVM Old-Age Pension Plan.........................................4
Pension System Performance before 1997 ..................................................5
Coverage and Total Payroll Tax ......................................................5
Labor Market Efficiency..................................................................6
Redistribution to the Elderly poor....................................................7
Savings-Insurance ............................................................................8
Financial Condition of the IMSS/IVCM Plan .................................9
Government Policy on Pension Reform ....................................................10
Redistribution toward the Poor Aged in the Long Run .................12
Savings-Insurance in the Long Run...............................................13
Recognition of Entitlements Accrued by Members in Mid-Career
........................................................................................................17
Human Resource Management: Needs and Portability .................18
Fiscal Policy Response ..............................................................................18
2. Bank Assistance to Mexico........................................................................................22
Genesis of Bank Involvement....................................................................22
Description of Bank Assistance .................................................................23
3. Impact of Bank Assistance ........................................................................................26
Outcome: Relevance of Objectives............................................................26
Mexican Development Needs ........................................................26
Bank Strategy.................................................................................27
Quality of Economic and Sector Work..........................................31
Outcome: Efficacy .........................................................................33
Institutional Development Impact..............................................................36
Sustainability..............................................................................................38
4. Results .........................................................................................................................40
Bank Performance......................................................................................40
Borrower Performance...............................................................................41
Other Players and a Counterfactual ...........................................................41
5. Lessons Learned and an Agenda for Future Action...............................................42
Appendix: Interviews in Mexico City (November 2002) .............................................43
References.........................................................................................................................45
Tables
1.1
2.1
Program Contribution Rates ...................................................................................11
Preliminary Rating of the FTAL.............................................................................24
i
Preface
This paper belongs to series of 19 country and regional case studies
commissioned as background research for the World Bank's Independent Evaluation
Group (IEG) report "Pension Reform and the Development of Pension Systems." The
findings are based on consultant missions to the country or region, interviews with
government, Bank, donor, and private sector representatives involved in the pension
reform, and analysis of relevant Bank and external documents.
This case study was authored by Salvador Valdés-Prieto in 2003. Salvador
Valdés-Prieto is a professor of economics at Catholic University of Chile and a
researcher at the Centro de Estudios Públicos, Santiago.
ii
1
1.
Background
Pension System Structure before 1997
1.1
Mexico’s pension system has undergone many changes through the years,
including significant reform measures in 1992 and 1997. The old Mexican social security
system (established in 1942–44) offered a rich menu of services for citizens later in life,
including savings for old age, longevity insurance, and survivors’ pensions. Institutions
that offered pension services were the Mexican Social Security Institute (IMSS), for
private-sector workers; the Institute for Security and Social Services for Government
Workers (ISSSTE), founded in 1959 through the unification of older plans for federal
workers and some state government employees; 32 public schemes serving state-level
government employees, and separate institutes for military personnel and the police,
university professors, and workers at several parastatals, including Petroleos Mexicanos,
and even for IMSS's own personnel, all of whom were exempt from contributing to
IMSS's general plan and to ISSSTE.
1.2
These are not specialized institutions because they also provide health insurance
and medical services (for active and pensioned members), workers compensation,
disability insurance, life insurance, maternity leave coverage (for example, loss-ofearnings insurance), child care services, and marriage bonuses. In addition, IMSS
provides social services, which include subsidized cultural and sport activities, vacation
resorts, public education about health and first aid, funeral services, and a host of other
activities. IMSS also provides (noncontributory) social assistance in the form of health
services for selected poor communities.1 Although the law endows each of these social
security institutions with a separate contribution rate, in practice, the resources are
pooled. For example, the Disability, Old Age, and Survivorship (IVCM)2 office at IMSS
had cash surpluses in 1995 and 1996, and these surpluses were used to finance the
delivery of medical services.
1.3
Until Mexico’s more recent reform in 1997, there were two branches of social
security, managed by separate institutions. The first branch, called the Retirement
Savings System (SAR), supervised by the National Commission of SAR (CONSAR),
was created in 1992 and began operations in January 1993. SAR required a two percent
contribution of declared earnings up to 25 minimum salaries from members of both IMSS
and ISSSTE.3 In exchange, the SAR offered a supplemental retirement benefit to
members when they reached age 65, became eligible for a pension under the IMSS or
ISSSTE plans, or upon receipt of an employer-provided pension. SAR did not mandate
1
See Articles 208 to 217 of the Social Insurance Law.
Old-age pensions will be referred to in this paper as the “IMSS/IVCM plan.” IVCM was a much-used
name for the following combination of programs within IMSS: pensions for old age and the unemployed
old, health insurance for pensioners, disability and survivorship insurance, and social assistance. Note that
the last two of these components are not considered part of old-age security in other countries. Health
insurance for pensioners is also managed separately in other countries.
3
Apparently, workers in states with separate plans, in Petroleos Mexicanos, and in the military were not
required to participate in the SAR.
2
2
longevity insurance because it allowed members to withdraw 100 percent of their benefits
as a lump sum when they reached eligibility. All of the funds collected by SAR were
invested at the central bank, and the law promised a real interest rate of two percent on
the funds.
1.4
Since 1972, the second branch of social security has offered subsidized loans for
home buying and construction to its contributors. The institutions in this branch are the
National Workers Housing Fund Institute (INFONAVIT), for covered private-sector
workers, and the National Institute for the Development of Housing for Government
Workers (FOVISSTE), for federal employees. The contribution rate to INFONAVIT is
five percent of declared earnings, up to a special ceiling, and the one for FOVISSTE is
six percent. Members who do not obtain a loan (or after they repay one and have reached
pension age) are allowed to use their account balances to buy a supplementary old-age
pension. Until 1992, however, the nominal interest rate on virtual account balances for
this purpose was zero percent. Given that actual inflation remained above 50 percent in
each of the seven years from 1982 to 1988, this meant a real return at below –33 percent
per annum, which drove the value of the accounts to nearly zero. In the 1992 reform, the
interest rate was required to be positive in real terms, thus creating a government
guarantee. The 1992 reform called INFONAVIT part of the retirement system and
created the "INFONAVIT sub-account,” whereby contributions are credited. In practice,
this is not managed as a retirement plan but, rather, as a housing subsidy program.
Redistribution toward the Elderly Poor
1.5
In the 1990s, Mexico set up successful redistribution programs for the rural poor,
such as Progresa (now called Oportunidades).4 However, Mexico did not provide
assistance pensions to the elderly poor on a noncontributory basis.5
1.6
Two provisions under the IMSS pension plan had progressive redistribution
toward the elderly poor as their specific aim and have been active since 1997. (IMSS is a
provider of health services to the rural poor, including the elderly poor.) One is the
progressive scale of replacement rates embedded in the benefits formula as a function of
the level of the base salary (expressed in multiples of the minimum salary). For a base
salary equal to 1.01 times the minimum salary and 9.6 years of contributions, the
promised replacement rate is 80 percent. For a base salary equal to the median—2.8 times
the minimum salary—achieving the same 80 percent replacement rate requires 34.9 years
of contributions.
4
Oportunidades is targeted at poor women with children, who must take instruction in particular
nutritional, health, and education services to become eligible. This information is from World Bank 2002,
p. 13, paragraph 34.
5
In 2000, Mayor of Mexico City Manuel Obrador, created the first universal noncontributory old-age
pension in Mexico, at Mex$ 600 /month (US$ 70 /month). The pension was available to the residents of his
city who requested it. Some ineligible residents in other towns tried to find a way to become recipients.
This may have been the harbinger of future social policies.
3
1.7
The second redistribution provision was the minimum guaranteed pension for
IMSS beneficiaries indexed by a 1989 law equal to 100 percent of the minimum salary. If
the minimum pension amount is larger than the pension set by the benefits formula plus
cost-of-living adjustments since inception, the government will pay the difference. The
amount set by the benefits formula does not depend on the balance of the INFONAVIT
sub-account.6
1.8
Access to these two redistribution provisions was limited by a further
requirement: completion of 500 weeks (9.6 years) of contributions. Moreover, for the
many workers who contribute fewer than 9.6 years in the formal sector, these subsidies
turned into a tax—no benefits whatsoever were paid and the member lost 100 percent of
the contributions. This implicit tax was regressive because it affected mostly women and
poor workers.
1.9
Progressive redistribution can also be achieved by an appropriate taxation
schedule. Just before the 1997 reform, the personal income tax had an exemption equal to
three minimum salaries. Also, the benefits paid by IMSS were exempt from income tax.
Savings-Insurance Programs
1.10 Delivering savings-insurance to the middle classes was another major aim of the
IMSS plan. The benefit formula is such that the pension amount rises proportionally to
the base salary, and pays higher pensions to higher-salaried members.7 The IMSS plan
offered defined benefits and financing through an unbalanced pay-as-you-go system, with
both workers and employers making contributions and other branches of social security at
IMSS absorbing the cash surpluses. The IMSS plan, however, was not supported by a
governance structure to help it meet savings-insurance goals at a minimum risk to
workers. For example, it did not include an independent office for long-term actuarial
projections and make them public.8 The IMSS is a unified federal organization
employing about 350,000 workers in 1992 and 371,186 in 2001, including those
delivering medical services.
1.11 Savings-insurance can also be offered to medium and high earners through tax
incentives for voluntary and occupational plans (third-pillar). Since 1993, the SAR
reform allowed favorable tax treatment for additional voluntary savings for old age by
individuals, but this attracted a very limited amount of funds.
6
See Article 168 of the Social Insurance Law; before the 1995–97 reform. When stipulating the pensions to
be supplemented by subsidies until the minimum amount is reached, it omits INFONAVIT benefits.
7
The same happens at ISSSTE, which offers a replacement rate of 100 percent upon completion of 30
years of service.
8
The first external and audited actuarial report of IMSS, required by a 2001 reform, was made public in
June 2002 and was a valuable first step. However, it excluded a valuation of the liabilities of the old-age
pension program because these were transferred to the federal government by the 1997 reform, as part of
the transition costs. See IMSS 2002, pp. ii and v.
4
Human Resources Management: Aims and Portability
1.12 Some large employers use contingent pension promises to permanent workers to
assure the recovery of the firm-specific training and team-building investments and to
select better employees. There is a tradeoff between this aim and the ease of transferring
pension rights between jobs under different mandatory plans. In Mexico, there were
about 8,000 employer-sponsored plans, covering a significant 20 percent of formal
employment. Most of these plans' assets—about five percent of gross domestic product
(GDP) in 1993—belonged to just 35 plans, sponsored by a few multinationals and
parastatal firms. The labor law set a minimum severance payment of 1.5 months of
salary per year of service, and the need to fund this liability led some employers to create
pension plans. The 1973 Social Insurance Law made some demands on employer plans,
and allowed IMSS to compete as a provider of services. Article 27 of this law allows
contributions to an employer plan to be excluded from the earnings base to which IMSS
contribution rates are applied (partial opting out),9 but only if the plan is authorized by the
Ministry of Finance on a case-by-case basis. Until 1997, however, there was no
regulatory agency in charge of supervising these plans. There are no regular statistics
about the types of plans or about their financial status. Thus, employer-based
supplemental plans did not benefit from a regulatory infrastructure that balanced the
interests of employers and employees.
1.13 The ISSSTE plan does not help the government-as-employer protect its
investments in personnel training because pension eligibility is not linked to a given
agency or career path. Therefore, it cannot help recover agency-specific training or
selection costs. The IMSS plan, ISSSTE plan, state-employee plans, and parastatal plans,
however, are not fully portable, as there is no system to allow workers that move among
the different sectors transferring pension rights from one plan to the other.
The IMSS/ICVM Old-Age Pension Plan
1.14 The IMSS/ICVM is a traditional single-pillar plan, in the sense that both
redistribution and savings-insurance aims are served together. Until 1992 the contribution
rate for the old-age pension was very low, as is common in pension plans that have not
yet matured. The contribution rate for old-age pensions was 3.0 percent of earnings. The
government contributed five percent of this, the worker contributed 25 percent, and the
employer contributed 70 percent. If the two percent for the SAR (fully paid by
employers, after INFONAVIT contributions) is added, the total contribution for old age
was less than five percent, which is way below contribution rates in mature, funded plans.
1.15 The pension age for the IMSS plan is 60 if the worker is fired or arranges to be
fired (old-age severance or Cesantía en Edad Avanzada), and 65 if the worker quits (old
age or Vejez). Members who continued working in their current job could not draw a
pension. Until 1997, however, a member was entitled to a (lower) pension with just 500
9
Opting out would be complete if contributions to the employer plan were deducted from the contributions
to IMSS.
5
weeks (9.6 years) of contributions. IMSS pensioners had an average contribution period
of 22 years, confirming that IMSS/ICVM was a plan that had not yet matured.
1.16 The benefit formula was unusually complicated because it depended on both
earnings and on the number of years of contributions or years of service (as in the United
States).10 In the range for median declared earnings, the following benefits formula
applied:
β2.76–3.00 = 0.3048 + 0.01958.(N – 9.6),
where β is the replacement rate and N is the number of years of contribution. The base
salary was the average of the last 250 weeks (4.8 years) of earnings, not adjusted for
inflation. Benefits for pensioners included a 13th pension each December (Aguinaldo).
After disbursement, the pension amount was adjusted for real price inflation once a year,
in February.
1.17 Pension benefits include health insurance coverage but only for health services
provided at IMSS facilities. The cost of this extra benefit was potentially very significant
in the long term, as international experience shows. An unusual aspect of IMSS/ICVM
health insurance for the elderly is that members earn the right to it by paying a
contribution of 1.5 percent of declared earnings when they are active, not when retired (as
in many countries). Most descriptions of the ICVM plan omit the fact that promised
benefits are not just pensions, but also include health insurance.
Pension System Performance before 1997
Coverage and Total Payroll Tax
1.18 In 1993, the IMSS plan had about 9.4 million contributors and about 1.4 million
pensioners. The analogous ISSSTE plan had about 1.5 million contributors and about
0.26 million pensioners in 1993, with an average declared salary and average benefits
much higher than at IMSS. The Petroleos Mexicanos plan had about 0.5 million
contributors, the military plan had about 0.3 million, and the IMSS-employee plan had
about 0.3 million. For the informal sector and self-employed (mostly in subsistence
agriculture), about 13 million workers were not covered by any plans. Mandatory plan
contributors were 40 percent of the employed (Ayala, 1997), of which IMSS covered 79
The benefits formula is a table of the formula: βk = ck+ak.(N– 9.6), where k indicates the range in which
the value of the base salary of the individual falls, expressed in multiples of the minimum salary. There
were 21 ranges that ran from (1.01–1.25) to (6.01–10) minimum salaries. For the highest earnings the
following benefit formula applied: β6.01–10 = 0.1300 + 0.0245(N– 9.6). Source: Article 167 of the 1973
Social Insurance Law.
10
6
percent of them. The declared taxable wage bill was about 7 percent of GDP11 in 1994,
which is low by international comparisons.
Labor Market Efficiency
1.19 The low coverage rate may have been partly because of large payroll taxes, which
may have created incentives to move production to the informal sector. The contrary
view is that payroll taxes were low compared with benefits. The total nominal tax rates
on private sector employment added up to an apparent 31.5 percent of declared earnings,
up to 10 minimum salaries, and to 14.5 percent on the surplus, up to 25 minimum
salaries,12 and zero thereafter. These rates need to be corrected, however, on three
accounts.
1.20 First, Article 27 of the Social Insurance Law requires contributions to
INFONAVIT to be excluded from the earnings base to which IMSS tax rates are applied.
Second, five percent of the remaining rate was a contribution to IMSS from general
revenue (the "government's contribution" in the tripartite scheme). A third correction is
required to take into account that 70 percent of IMSS contributions were paid by the
employer and only 20 percent by the worker. In contrast, the SAR and the INFONAVIT
rates were paid at 100 percent by the employer. Therefore, the total effective rate as a
proportion of the employer’s cost, equivalent to the case in which contributions are paid
entirely by workers, was 24.4 percent.13 This rate is modest when compared with other
countries in the Organisation for Economic Co-operation and Development (OECD),
given the volume of insurance and other services promised in exchange. However, it may
still be too high for Mexico’s conditions, where uncovered employment is 60 percent,
and where the real wage elasticity of labor to the covered sector is likely to be higher than
in the OECD.
1.21 The benefits-contributions link at the individual level was extremely weak in
Mexico, so this tax rate was likely to be perceived as a pure tax on declared earnings, to a
larger extent than in other OECD countries. The five percent contribution rate for housing
(INFONAVIT) sets Mexico apart from other countries, and moreover, is perceived as a
pure tax because of the way in which housing benefits are distributed: subsidized loans
are distributed mostly to members of powerful unions and the subsidy is not proportional
to past contributions. Because contributions to INFONAVIT, prior to 1990, were fully
taxed away by inflation, there is little trust in a benefits-contributions link. Regarding
11
This number is estimated by multiplying 9.4 million contributors by an average declared salary of 2.8,
multiplied by a minimum salary of 80 dollars per month, multiplied by the exchange rate of 3.38 for 1994,
multiplied by 12 months, and dividing by the GDP for 1994, which was 1,226 billion pesos.
12
The ceiling for contributions to SAR (2 percent contribution) and to the health portion of IMSS (12.5
percent contribution) is 25 minimum salaries.
13
Valdés Prieto 2002, Chapter 5, shows that the equivalent total tax rate on workers is (θE + θL)/(1 + θE)
in the absence of INFONAVIT and SAR taxes, where θE is the employer's contribution rate and θL is the
worker's contribution rate. In the presence of INFONAVIT and SAR taxes, the equivalent total tax rate is
[5.0 + 2.0 + (31.5 -5.0 -2.0) x (1-0.05)] / (1 + 0.02 + 0.05 + 0.70 x (0.315- 0.05-0.02)) = 24.39 percent.
7
SAR, the funds were fully invested in government securities and paid a two percent real
return, which was miserly compared with market returns.
1.22 The amount of health services provided by IMSS was unrelated at the margin to
the individual's money contribution, so the marginal tax rate was 100 percent. As for
IMSS disability and life insurance, its benefits were earnings related, so the link was
acceptable. As for the IMSS’s old-age pension promises, those earning the minimum
salary meant that, at the margin, one more year of contribution did not raise the
replacement rate.14 For those earning the median salary, one more year of contribution
raised the replacement rate by two percentage points, which is way below the increase
afforded by market interest rates. In addition, this promise was subject to inflation taxes
and to the (political) risk that the formula would be changed. Moreover, the short
averaging period (4.8 years) used to determine base salary meant that contributions made
before that period had zero marginal effect on pension benefits, cutting the benefitscontributions link to zero in those periods.15 In summary, the overall benefitscontributions linkage was nearly zero, allowing low coverage to be explained in part by
the 24.4 percent payroll tax rate.
Redistribution to the Elderly poor
1.23 Some redistribution provisions were successful and others failed. The contribution
requirement excluded most of the rural poor, who have the shortest periods of
employment in covered jobs. For this same reason, the implicit 100 percent tax on those
members not completing 500 weeks of contribution was highly regressive. Old-age
benefits covered just 20 percent of the urban elderly16 and probably a negligible
percentage of rural elderly. Thus, the urban coverage rate for benefits was a mere onehalf of the coverage rate of contributions, reversing the situation in other countries where
the coverage rate for benefits is higher thanks to noncontributory pensions. On the other
hand, the minimum pension was successful, in the sense that 80 percent of IMSS
pensioners had their calculated pensions topped-up to reach the minimum by this
program.17 The progressive benefits formula at IMSS can, at most, redistribute wealth
among plan members, excluding the relatively better-paid employees at the large
parastatals, the federal and state governments, the higher-income self-employed, and the
owners of capital. This situation could not be reversed by a government contribution of
just five percent of the IMSS rate, which meant just 0.98 percent of the employer's cost.18
14
The subsidy paid by the minimum pension was withdrawn rather harshly when the average of declared
earnings surpassed the minimum salary by just 1 percent, because the replacement rate fell from 100
percent to 80 percent (provided the member had contributed for 9.6 years). This implies an infinite
marginal tax rate at that point.
15
The short averaging period also stimulated over-reporting the last 5 years of earnings, but did not
stimulate overtime hours of work because their remuneration was excluded from taxable earnings by
Article 27 of the 1973 Social Insurance Law.
16
This is taken from World Bank 2002, p. 13, paragraph 34.
17
Ayala, 1997.
18
The government contribution was just [(31.5 – 5.0 – 2.0) x (0.05)] / (1 + 0.02 + 0.05 + 0.70 x (0.315 –
0.05 – 0.02)) = 0.9867 percent of employer costs.
8
Redistribution through IMSS was also limited by the ceiling on taxable earnings for
pension purposes, set at 10 minimum salaries. Finally, a benefits formula that sets the
base salary at the average of the last 4.8 years of earnings redistributes toward those with
steeper age-earnings profiles, which are not the poor.
Savings-Insurance
1.24 The provision of a savings-insurance vehicle for the middle classes was a failure.
The promised replacement rate (40–80 percent plus health benefits) was generous
compared with the apparent contribution rate (6.5 percent), in the sense that the implicit
promised rate of return was large in real terms.19 However, the IMSS defaulted on this
promise because of repeated policy decisions to avoid implementing indexation rules to
protect the real value of pension promises. Inflation was also never below 50 percent per
annum in 1982–88.20 Indexation of the minimum pension occurred after the high
inflation of the 1980s wiped out the value of most pensions.
1.25 The base salary is the average of nominal covered earnings during the last 250
weeks (4.8 years) before the pension is issued and, therefore, the pension's real value
when it is first issued is a negative function of the inflation rate. Inflation indexation for
paying pensions occurs at a fixed regular interval (in February), so a permanent increase
in the inflation rate reduces the average real value over an annual cycle. The high
proportion of pensioners on the minimum pension proves how much the real value of
pensions was devalued during the inflationary period 1982–88. Inflation and willful
omission of indexation clauses, therefore, destroyed the reputation of the IMSS/IVCM
plan as a savings-insurance vehicle for the middle classes. Although this default on
pension promises could also have allowed the government to reduce its deficit and
stabilize the economy and recover growth, it did not happen in 1982–87. Administrative
costs at IMSS were huge, owing to the excessive compensation and staffing agreed to
with the labor unions.
1.26 A retirement saving scheme, SAR, was introduced in 1993. But the SAR program
was poorly managed, as neither CONSAR, the regulatory agency, nor private banks that
collected contributions successfully addressed operational problems, and members began
to accumulate multiple accounts when they switched employers. In addition, the SAR's
failure to mandate longevity insurance— as benefits could be taken as a lump sum—
making the program was ineffective in protecting improvident members in old age.
1.27 Old-age pensions were offered as in either defined-benefit or defined-contribution
form, with different risk structures respectively. The IMSS/IVCM was a defined benefit
plan, while the SAR's benefit was defined contribution, unless the member chose to buy
19
However, as the marginal real interest rate in the financial market was also very large in most years
during 1982–95, the presence of planned redistribution toward the earlier generations of members is not
evident.
20
The Mexican CPI inflation rate stayed above 50 percent per annum from 1982 to 1988, peaking at 159.2
percent in 1987. In 1989–94 it fell steeply to 7.0 percent, but in 1995 it rose again to 52.0 percent. Inflation
was 27.7 percent in 1996 and 15.7 percent in 1997.
9
an annuity in a private insurance company—a defined benefit. INFONAVIT benefits
were defined contributions if paid in a lump sum. Even IMSS/IVCM annuities were, in
effect, defined contribution benefits with respect to inflation risk, which was not covered.
Financial Condition of the IMSS/IVCM Plan
Expenditures on IMSS/IVCM were very low in the early 1990s, about 0.40 percent of
GDP. This figure includes expenditures on disability and survivor pensions, social
assistance, and health insurance for pensioners because they are all part of IVCM. On the
other hand, IVCM revenue was 8.5 percent of taxable earnings, including a contribution
for disability and life insurance,21 and this was about 0.60 percent of GDP.22 Thus ICVM
had a cash surplus of about 0.20 percent of GDP in 1994, which was projected to fall
gradually and disappear by 2007.
1.28 Government projections showed that IVCM was not sustainable as cash flows
turned substantially negative in the long run. To interpret such figures, it is essential to
distinguish three situations: (i) balanced pay-as-you-go finance under steady-state
demographics, which is financially stable because cash flows are zero every period. In
this (theoretical) case, accrued liabilities are equal to the (positive) difference between the
present discounted value of future contributions and the present discounted value of
future benefits; (ii) an initially balanced pay-as-you-go finance that suffers an aging
shock, for example, an increase in the portion of the population above age 60 (for
example, in Mexico, from 6.4 percent in 1990 to 10.4 percent in 2020) will exhibit
negative cash flows until the parameters are adjusted; and (iii) under many plans the
actuarial deficits go over and above the levels that can be explained by an aging shock,
because of a combination of an immature system with generous vesting for initial
generations, and a preference of politicians to delay parameter adjustments. Mexico's
actuarial deficits combined the second and third situations.
1.29 Mexico did not have independent institutions in charge of publishing actuarial
projections for IMSS/IVCM on a regular basis. The figures cited below are from a private
actuarial office hired by IMSS in mid-1995, and from a model developed by the Ministry
of Finance. The parameters of both models were not reported in World Bank documents
and the annual cash flow projections were not published until 1998. The government
reported in 1995 that the IVCM/IMSS plan had an actuarial deficit of 73.4 percent of
GDP for those generations that had joined the IVCM/IMSS as of the valuation date.
1.30 The breakdown of this figure by generations is informative. First, the liability to
the generation already pensioned was just 7.62 percent of GDP, in line with the modest
benefit expenditure of 0.40 percent of GDP for 1994 inherited from the 1982–88
inflation. Second, the present value of expected future contributions from currently active
21
The 8.5 percent excludes the 2 percent for SAR. The figure consists of 3.0 percent for old age + 1.5
percent for old-age health + 3.0 percent for disability and death + 0.4 percent for social assistance + 0.6
percent for administrative expenses = 8.5 percent.
22
This is obtained by multiplying a taxable wage bill of 7 percent of GDP by 8.5 percent, which is the
combined IVCM contribution rate.
10
members was 14.12 percent of GDP, in line with low coverage, low declared earnings,
and low contribution rates. Third, the present value of expected future benefits to
currently active members was 79.95 percent of GDP, way above the other numbers and
driving the total. This large value is a result of the following: (i) an assumption that
inflation would fall in Mexico, thus allowing the generosity of the benefits formula to
emerge (a pension age of 60, a 100 percent replacement rate for those on low earnings,
and a vesting period of 9.6 years); (ii) the fact that benefits include health insurance for
pensioners, and its cost grows in real terms; (iii) economic growth, so new pensions
would be based on the average of 4.8 years of much higher real earnings; and (iv) aging
members. The IVCM/IMSS plan would thus become insolvent without new legislation.
1.31 Annual cash flow projections were not available during the reform debate.
According to projections later done by Sales, Solís, and Villagómez (1998), IMSS/IVCM
expenditures for the old system were 1.55 percent of GDP for 1997, 5.43 percent of GDP
for 2022–26, and 14.01 percent of GDP for 2047, at 3 percent annual GDP growth. These
amounts are very sensitive to growth assumptions because the last projection dropped to
6.29 percent of GDP when using five percent annual growth.
1.32 The financial situations of other mandatory pension plans in Mexico are worse,
relative to the number of members. The actuarial deficit at ISSSTE was about 15 percent
of GDP, and state employee plans and parastatals were believed to add another 35
percent of GDP in actuarial deficits (Ayala, 1997). The IMSS employee plan had an
actuarial deficit of 5.2 percent of GDP as of 2001. If potential government liabilities in
private sector occupational plans are ignored, total hidden pension debts in Mexico
before the reforms can be estimated at 130 percent of GDP.
Government Policy on Pension Reform
1.33 Mexico implemented a pension system reform on July 1, 1997 established by two
laws, dated December 1995 and April 1996.23 Institutionally, the reform split up the
IVCM combination of programs, leaving IMSS with health insurance for pensioners,
disability and survivorship insurance, and social assistance. The contribution for
disability and survivorship insurance was cut from 3.0 percent to 2.5 percent of earnings,
because benefit promises were cut;24 the 0.4 percent contribution for social assistance
was replaced by a transfer from the treasury; and the 0.6 percent contribution for IMSS's
administrative expenses was redirected to the elderly (see table 1.1). Old-age pensions
and old-age severance were taken from IMSS and merged with the SAR retirement
pension into a new old-age pension system. The new system is fully funded (Sociedades
de Invesión Especializadas en Fondos para el Retiro), and contrary to SAR, the funds
are, in fact, managed by private firms called Administradoras de Fondos para el Retiro
23
The original reform date of January 1, 1997, was delayed to assure effective implementation.
The eligibility requirement for disability insurance was raised dramatically, from 150 to 250 weekly
contributions (from 2.9 to 4.8 years of contribution). This change excluded those disabled young workers
with longer periods of work outside the covered private sector (40 percent of employment) from receiving
benefits.
24
11
(AFORE). These firms may be established by unions, private enterprises (including those
that are foreign owned), and IMSS. INFONAVIT continued distributing housing
subsidies, as before, except for some limited reforms to its management.
Table 1.1: Program Contribution Rates
Program name before
reform
SAR: Retirement
account at the central
bank.
Before reform
on July 1, 1997
2.0 %
After reform on July
1, 1997
2.0 %
Program name
after reform
AFORE:
Retirement sub-accounta
n/a
---
Old age and severanceb
3.0 %c
Social assistanceb
0.4 % c
Administrative
expensesb
0.6 % c
Disability and survivors
insuranceb
3.0 % c
Health insurance for
pensionersb
1.5 % c
Health and maternity
insuranceb
12.5 %
Child care serviceb
Work accident
insuranceb
INFONAVIT
Housing subaccount
1.0 %
US$ 4.4 (in 1997) per Social quota --new transfer
contribution
from Treasurya
Old age and severance sub4.5 % c
accountsa
Added to old age and
Financed with a new transfer
severance subfrom the treasury
accounts
Added to old age and
Eliminated
severance subaccounts
2.5 % c
0.5 points moved to Disability and survivors
old age and
insuranceb
severance
Health insurance for
1.5 % c
pensionersb
US$11.1/contribution
(by employer)
Health and maternity
insuranceb
+ 8 % of
+ new transfer from Treasury
(E-3MS)
(25 % worker and 75 of US$ 11.1/mo. per worker
% employer)
1.0 %
Child care serviceb
2.5 %
2.5 %
Work accident insuranceb
5.0 %
5.0 %
INFONAVIT
Housing sub-accounta
Note: The contribution base and the ceilings on taxable earnings differ among SAR, IMSS, and INFONAVIT.
a. AFORE.
b. IMSS.
c. Figures include a 5 percent government share of the contribution rate. Workers have a 25 percent share and employers have a 70
percent share, and the sum of their rates is 95 percent of the reported rate. The contribution rates to the AFORE system reported
here are before fees are subtracted.
1.34 This reform covered only members of IMSS, that is, private-sector workers in
covered firms, excluding ISSSTE members, workers at parastatals and state governments,
and the uncovered private sector. Across generations, those pensioned at IMSS before
1997 retained their pension rights, although the obligation to pay them was shifted from
the IMSS to the Treasury, while IMSS retained the benefits payment function. Those
joining covered employment for the first time after 1996 were prohibited from belonging
to the IMSS regime for old-age pensions. Benefit rights accrued by all generations of
active IMSS members as of 1996 were recognized (discussed below).
12
1.35 The1997 reform took collection away from the private banks and returned it to
IMSS, which continued collecting other for programs as well. A newly centralized
"information collection and database management" function was created. A private
monopoly was chosen to operate it through international bidding, with tariffs set by
CONSAR.25 The reform introduced a new social security number, called the Clave Unica
de Registro Poblacional (CURP) and its 18 characters include data on the member’s age,
sex, and birthplace.
Redistribution toward the Poor Aged in the Long Run
1.36 The reform replaced the progressive benefit formula at IMSS with a flat universal
subsidy from the Treasury for each contribution or "social quota." These were paid into a
member's saving account for old age, even those with high earnings. The amount of this
benefit was set at 5.5 percent of the minimum salary (US$ 4.4 per monthly contribution,
as of 1996, and about US$ 7.0 per contribution, as of early 2004). Over the years this
amount will decline relative to wages because for this purpose the minimum salary was
indexed to price levels (discussed below). Members who spent longer periods in informal
jobs received the social quota less frequently than those with more periods of
contribution.
1.37 Members with an incomplete contribution records no longer suffered from a 100
percent tax on past contributions if the number of contributions was below 500 weeks.
These members were allowed to withdraw their accumulated funds, including interest, as
a single lump sum at age 60.
1.38 The reform raised the eligibility requirement for the minimum pension guarantee
from 500 to 1,250 weeks of contributions and redefined a "week" from a calendar week
to seven full working days. This meant an increase from 9.6 to 33.7 years of contributions
(not 25 years, as was widely reported). Moreover, indexation of the minimum pension
guarantee was changed from wages to prices (a reduction in effect). The old minimum
pension was 100 percent of the current minimum salary, but the new one was 100 percent
of the minimum salary as of December 31, 1996, and it was adjusted for inflation using
the Consumer Price Index (CPI). Minimum pensions were also withdrawn if a pensioned
member found covered employment again with a different employer.26
1.39 The amount of subsidy implicit in any minimum pension is the difference
between the minimum amount and the self-financed pension. The 1997 reform changed
the applicable self-financed pension for the generations that entered the labor market after
1997, thus modifying their subsidy. Originally, the self-financed pension was the IMSS
benefit alone, excluding INFONAVIT benefits. The reform redefined the applicable selffinanced pension in two opposite ways: (i) increasing old-age benefits by financing with
25
The winner in the bidding contest was PROCESAR. Its main function is to collect individualized
information on worker contributions from employers, validate it, and pass it on to AFOREs to update the
individual accounts.
26
Compare Article 173 of the new Social Insurance Law with Articles 174 and 123 of the old law.
13
the social quota and INFONAVIT contributions, and also (ii) decreasing old-age benefits
because they were likely to be below the generous pre-1997 IMSS benefit level for most
members.
1.40 The government created a new program that extended a flat health subsidy, equal
to 13.9 percent of a minimum salary, to self-employed poor workers (and their families),
who also contributed a flat rate of 22.5 percent of a minimum wage on their own.
Savings-Insurance in the Long Run
1.41 The total contribution rates for IVCM were not increased, despite their low
initial level. However, the following contributions were diverted to old-age savings: 0.5
percent originally designated for disability and survivorship insurance, 0.4 percent
originally designated for social assistance, and 0.6 points originally designated for
IMSS administrative expenses. Including the SAR, total contributions for old-age
savings reached 6.5 percent of earnings, which is still modest by international
standards, especially because AFORE commissions are subtracted at an average rate of
about 1.6 percent of earnings,27 leaving just 4.9 percent to accumulate. The five percent
paid to INFONAVIT is frequently described as part of the old-age pension system, but
this characterization is inaccurate because it is a housing subsidy program, so it is
uncertain whether it will live up to its promised zero percent real return. The social
quota is sometimes reported as part of the contribution rate, but this is confusing
because the quota will fall over time relative to average earnings.
1.42 To estimate the replacement rate for the average male earner, consider a simple
simulation. Assume that a total flow of 11.9 percent of the 1996 earnings of an average
earner obtains the following rates of return: a five percent real rate of return on net
contributions and on the social quota contribution (the rates of return earned by Mexican
pension funds are higher); and a one percent real rate of return on the INFONAVIT
contribution. Also assume that the internal rate of return paid by joint annuities in the
dispersal phase is four percent real, that the growth rate of an individual's salary is three
percent real (including a premium for experience) until age 40 (and zero thereafter), that
the rate of contributions is 50 percent (in line with coverage of 40 percent), and that the
duration of retirement 25 years. The result is a final-salary replacement rate of 27.5
percent for the average earner who begins to receive a pension in 2037. This is low, but if
the rate of contributions rises to just 75 percent, the replacement rate increases to 41.2
percent, which is acceptable. Bank staff focused on the possibility of raising
INFONAVIT returns to equal those of the pension funds, a case where the replacement
rate rises to 36.9 percent even if the contribution rate stays at 50 percent. Lower rates of
return of four percent for the active phase, and three percent for annuities, compensated
for with an increase in INFONAVIT returns to four percent real and a 75 percent
contribution rate, leaving the replacement rate at 35.6 percent. Note also that health
insurance benefits in old age should be added for members that contribute 1.5 percent to
27
Approximate average for June 1998. As of January 2004, the commission rate as a percentage of salary at
BANAMEX, the largest AFORE by members, was the same 1.70 percent observed in June 1998.
14
IMSS during 20.2 calendar years. Unlike the old system, these replacement rates are
financed on a sound basis.
1.43 The standard pension ages—60 for those who have been dismissed and 65 for
those who quit—were not modified, but three significant changes were adopted. First, the
required number of contributions for those starting a pension (either an annuity or a
scheduled withdrawal) was raised from 500 to 1,250 weeks, with weeks redefined as
seven working days (9.6 to 33.7 calendar years). Second, a new benefit was created for
members aged 60 (dismissed) or 65 (quit) that did not meet this contribution requirement,
which was most—the payment in a single lump sum of 100 percent of the amount
accumulated in the individual's account.28 Third, an early benefit was created—members
were allowed to take a pension as soon as the total account balance was enough to
finance an immediate pension equal to 130 percent of the new minimum pension (as of
December 1996 indexed to prices) plus proportional survivorship benefits free of age or
contribution requirements.29 The early benefit had to be taken as an annuity of at least
130 percent of the minimum pension, but the surplus could be a lump sum.
1.44 The mandate to contribute was weakened by the low eligibility requirements in
the early pension: effectively, members were allowed to stop contributing30 as soon as the
AFORE account balance was enough to finance 130 percent of the minimum pension. At
age 35, this could mean an account balance of US$ 26,000 for a member who is single.31
Because this amount is indexed to prices, it will fall relative to average earnings.
Moreover, because voluntary contributions are also allowed, a member who only has, for
example, US$ 20,000 in his AFORE account and declares earnings equal to six times the
minimum salary, can circumvent the mandate and save for old age at his own pace by
doing the following: borrow US$6,000 from a bank; deposit this in the AFORE account;
ask for an early pension; stop contributing for old age, health, and to INFONAVIT; use
the cut in contributions for US$1,400 per year32 to pay off the bank loan in four years; and
be free of all contributions and AFORE fees thereafter.
28
See Articles 154 and 162 of the Social Insurance Law. This includes the balance at the INFONAVIT
sub-account.
29
See Article 158 of the Social Insurance Law.
30
More precisely, if the pensioned member continues in a covered job, the employer must deduct the
contributions and the member has to claim these back from the AFORE periodically. The returned
contributions are subject to income taxes. The tax impact is the same as being exempt from mandatory
contributions.
31
The monthly pension to be financed is US$ 80 x 1.3 = US$ 104 at the December 1996 exchange rate. A
real interest rate of 4 percent per annum was assumed. If the member is married and/or has children, then
additional savings are required to finance survivor pensions.
32
This amount is estimated on the basis of taxable earnings of US$ 480 per month in terms of the 1996
parameters. The contributions avoided by both worker and employer are a 95 percent share of the 4.5
percent for old age and severance, 100 percent of the 2 percent for retirement, 100 percent of the 5 percent
for INFONAVIT, the full contribution for health insurance of 8 percent (in 1997) in excess over three
minimum salaries, the full flat employer contribution of 13.9 percent of a minimum salary (US$
11.1/contrib. in 1997 $), the full 3.5 percent of contributions to IMSS for child care and work accident
insurance, and 95 percent of the 4.0 percent for IMSS, pensioner health insurance, and disability and life
15
1.45 This type of arbitrage is not prevalent in Mexico. One explanation is that a large
part of the US$1,400 savings cut employers' contributions, so the member would have to
bargain for a higher salary in order to recapture that part. Another is that once pensioned,
the member stops receiving the flat social quota (US$4.4 per month in 1996). A third
explanation is that high earners get a tax incentive for mandatory savings. A fourth is that
some members value the insurance lost and dislike a pension amount that would be just
15 percent of earnings (although voluntary savings could make up for that). Health
insurance is granted to pensioners only if they have 750 weeks (20.2 years) of
contributions,33 but as health services are delivered mainly at IMSS hospitals, quality may
be too low to be acceptable to a worker at this income level. Another likely explanation is
the absence of information.
1.46 The annuitization mandate was also reduced, by increasing the size of the lumpsum benefit available at old age, while reducing the pension benefit. The old lump-sum
benefit was 100 percent of the amount accumulated by the SAR's two percent
contribution rate. The new lump-sum benefit is the surplus accumulated with the 6.5
percent contribution rate to AFOREs (minus fees), plus the account balance at
INFONAVIT, plus the accumulation from the Social quota, minus the amount needed to
finance a pension equal to 130 percent of the 1996 minimum pension, plus proportional
survivorship benefits (both annuities are indexed to prices, so they fall over time relative
to average earnings). In addition, 100 percent of the amount preserved to purchase a
"pension" can be invested in a scheduled withdrawal, which is different from an annuity
and does not provide individual longevity insurance.
1.47 Another innovation in the Mexican reform was partial liquidity for mandatory
savings at an AFORE, called "Help for the Unemployed" (Ayuda de desempleo).
Members were entitled to withdraw a lump sum equal to the lesser amount of either 10
percent of the sub-account balance or 65 days of average daily salary (three monthly
salaries). The eligibility requirements are: (i) no contributions for 46 days (two calendar
months) since being dismissed from a covered job; and (ii) no withdrawals during the last
five calendar years.34 In 2002, the aggregate flow amount withdrawn owing to this
benefit, plus the marriage benefit, was about 1 percent of the contributions to AFORE,
net of commissions.
1.48 Each AFORE was allowed, in principle, to offer several different portfolios to
each member. However, a regulation required a single portfolio initially and it was still in
insurance. The savings are 0.95 x (4.5 + 4.0) + 2 percent + 5 percent + 3.5 percent + 11.1 + 8 percent x (63) / 6 per month = (11.1 + 480 x 0.227125) x 12 = US$ 1,410/year.
33
According to Articles 154 and 162 of the Social Insurance Law.
34
See Article 191-II of the Social Insurance Law. However, Article 198 of the Social Insurance Law
creates a penalty at the margin for withdrawing funds because the number of contributed weeks is reduced
in proportion to the amount of funds withdrawn, making it less likely to become entitled to the minimumpension guarantee. Apparently a subsequent reform (possibly in 2002) raised one of the benefit options
from 65 to 75 days of earnings, but a further eligibility requirement was added—250 weeks (6.7 calendar
years) of contributions.
16
force as of early 2004.35 Portfolio variety among different AFOREs is small, partly
owing to very tight regulations on portfolio composition. International investment is
prohibited, investment in domestic shares is prohibited, and at least 50 percent of the
portfolio is required to be invested in CPI-indexed bonds, which up until 2004 was issued
mostly by the government. In the long term, the new system will follow a defined
contribution rule for the accumulation phase. The benefits taken as pensions can be real
annuities issued by private life insurance companies (defined benefit) or scheduled
withdrawals (defined contribution).
1.49 The incentives for members to choose the AFORE with the lowest commissions
are dampened by the fact that no AFORE fee is deducted from take-home wages. Instead,
the fee is deducted from benefits, which are far off in the future and thus seem
unimportant to many members. This incentive is dampened also by the complexity of the
system just described and the variety of fee schedules. However, the reform did allow
CONSAR to decide how to allocate new members who had clearly not chosen an
AFORE—about 200,000 per month.36 Since 2000, CONSAR has exercised this option in
a way akin to a bidding contest,37 favoring the two AFOREs with the lowest
commissions. This created a new way for firms to enter the industry, with the first entries
into the Latin America AFORE business in 2003—Azteca and Actinver. Their
presence—although still modest—created incentives to cut commission rates, and six
AFORE did so as of February 2004. The cut in commission revenue, however, is still
modest in the aggregate.
1.50 The reform significantly reduced the potential fiscal costs of health insurance for
pensioners. It introduced a contribution requirement for entitlement at 750 weeks (20.2
calendar years) without reducing the amount of the contribution for pensioners’ health
insurance.
1.51 The active worker’s contribution for health insurance was cut by about five
percentage points. The original contribution rate was 12.5 percent of covered wages, of
which 95 percent was paid by workers and employers. The new system included the
following: (i) a flat employer contribution, set at 13.9 percent of a minimum salary (US$
11.1 per month); (ii) a contribution of eight percent of the portion of each covered wage
that exceeded 3 minimum salaries (in 1997, of this eight percent rate, two percentage
points were paid by the worker and the rest was paid by the employer);38 and (iii) a new
budget contribution, from general revenue, equal to a flat 13.9 percent of a minimum
35
In 2002, a second portfolio was allowed but only for voluntary tax-favored savings. Only three AFOREs
offer it and the take-up rate is low.
36
Power granted to CONSAR by Article 17-transitory of the Social Insurance Law.
37
However, undecided workers are relatively young (low numbers of contributions) and have lower
individual account balances as compared with other contributors. The fees they pay are modest. Azteca had
losses in 2003.
38
See Article 106 of the Social Insurance Law. According to Article 19-transitory, these tax rates will be
modified gradually over a 10-year period until the following steady-state scheme is reached: a flat
employer contribution set at 20.4 percent of a minimum salary, plus the flat contribution from general
revenue equal to 13.9 percent of a minimum salary, plus 1.5 percent of the portion of each covered wage
that exceeds 3 minimum salaries. Of this 1.5 percent, 25 percent is paid by the worker.
17
salary (US$ 11.1 per month). This reform maintained IMSS's revenue constant but cut
contribution rates.39
Recognition of Entitlements Accrued by Members in Mid-Career
1.52 Entitlement recognition was done by guaranteeing that each transition member
(those who had joined the plan before January 1, 1997) would be free to choose between
the benefit packages offered by the new and the old systems, known as the “lifetime
switch option.” The reason the government offered this option was because the
Constitution had established that the right to property greater than the promised benefit
becomes fully vested when one becomes a plan member, not when the pension is issued,
as under other plans. This point was later challenged.
1.53 It is important to make clear the differences between the new and old systems.
The new system's package includes: (a) a lump sum plus a pension, or a lump sum alone,
financed with the balance at the AFORE; (b) the SAR lump sum; and (c) the
INFONAVIT benefit. Entitlement is given to the new minimum pension guarantee,
depending on new eligibility requirements, which are very tight but encompasses benefits
(a) + (b) + (c).
1.54 The old system's package includes: (i) the IMSS annuity the member would have
been entitled to in the absence of reform, including the old minimum pension guarantee
with old eligibility requirements; (ii) the SAR lump sum; and (iii) the INFONAVIT
benefit.40
1.55 When a transition member chooses the old benefit formula, his entire
accumulation is paid in one lump sum to the Treasury, which issues the IMSS annuity in
exchange. In present-value terms, the government is worse off but it receives a cash
infusion. In the other case, where the transition member chooses the AFORE package, the
government is better off in present-value terms but its cash position does not improve.
For example, it is highly likely that the annuity equivalent of the sum of benefits offered
through the old system will be higher for someone claiming benefits in 2007, provided
inflation remains low, as in 2000–03. However, many transition members may prefer the
new benefit package because there is more in a lump sum, even if its present value at
market interest rates is lower.
1.56 For transition members that choose the old system's benefit package, the
government will continue to bear approximately 70 percent of the investment risk of the
individual account balance managed by the AFORE, with a strike price given by the
39
It is known that average earnings were 2.8 times the minimum salary. As an example, assume that only
40 percent of contributors have earnings that exceed 3 minimum salaries, and that the average of such
earnings, up to the ceiling, is 5.03 minimum salaries. Then the new system for health collects the same as a
tax rate of [13.9 percent + 13.9 percent + 8 percent . (5.03-3) . 0,40] / 2.8 = 12.25 percent, which is the 95
percent portion of the original tax rate paid by workers and employers. The other 5 percent of the tax rate
was paid by the government.
40
See Articles 3-transitory and 13-transitory of the Social Insurance Law.
18
IMSS annuity. This figure is not 100 percent because the 2.0 percent contribution for the
SAR sub-account is not replaced by the IMSS annuity but, rather, is paid to the transition
member that claims benefits. The government will also share 70 percent of the
commissions charged by the AFORE, for the same reason. This is separate from the
investment risk guarantee of the INFONAVIT sub-account, at zero percent real return.
The government will also bear the aggregate longevity risk, proportional to the volume of
IMSS annuities chosen by the transition members. The government does not bear
investment risk for transition members that choose the new system's benefit package
(except for the new minimum pension, which is negligible).
1.57 The presence of this 70 percent investment guarantee on the returns of the new
pension funds means that sophisticated transition members choosing the old benefit
package have an incentive to prefer the most risky investment strategies available.
However, many transition members who do not plan to complete 33.7 calendar years of
contributions—because doing so would prevent them from choosing all benefits as lump
sums under the new system—do not suffer distortions in their preferences regarding
investment risk.
1.58 The incentive to choose the AFORE with the lowest commission is muted for
transition members who are entitled to the 70 percent investment guarantee on net
returns.
Human Resource Management: Needs and Portability
1.59 The 1997 reform gave CONSAR new power to supervise and regulate employer
plans. CONSAR established a voluntary registration system, with the incentive that
members of registered plans would be allowed to transfer their account balance at the
AFORE to their private plans, provided the latter pay a sufficient pension. Apart from
this, the reform created two tax-favored vehicles for voluntary individual savings for old
age:41 (i) the option to contribute more than 6.5 percent of earnings to the old-age
account; and (ii) contributions to a separate voluntary account at an AFORE—but not at
banks, insurance companies, or mutual funds—which could be withdrawn at least once
every six months.
Fiscal Policy Response
1.60 Several sources have estimated total government expenditure related to the
IMSS/AFORE pension programs from 1997 including: Cerda and Grandolini 1997;
Ayala 1997; and Sales, Solís, and Villagómez 1998. This accounting concept is very
useful for budget programming, but it is not directly related to the evaluation of a pension
reform. Several expenditure programs existed before the reform and this approach does
not identify the differences made by the reform. The "total cost" for 1997–98 has been
variously estimated at one percent of GDP (Cerda and Grandolini 1997), at 0.931 percent
of GDP (Ayala 1997), and at 0.48 percent of GDP (Sales, Solís, and Villagómez 1998).
41
See Articles 191-I and 192 of the Social Insurance Law.
19
1.61 The Mexican reform diverted contribution flows away from IMSS and from the
SAR account at the Central Bank to the new pension funds, net of AFORE commissions,
at about 4.9 percent of covered earnings, or 0.343 percent of GDP in 1997. This is the
"transition deficit" created by the shift to funding of the second pillar.42 If this side of the
budget equation is estimated, it is not necessary to project the other side, and it does not
matter if the cash flow under the initial plan was different from zero. Payments to those
already pensioned as of the reform date and the benefits to be paid to the transition
generation would have occurred anyway, with or without the shift to funding, given the
lifetime switch option. Similarly, the five percent government share in contributions
would continue, with or without the shift to funding.43 The Social quota is a new program
that poured about 0.143 percent of GDP into the new pension funds. This additional
"transition deficit" was created by shifting the first pillar to funding.
1.62 The AFORE commission, a new program, diverted 1.6 percentage points of
earnings (0.112 percent of GDP) away from the general government, and was not
accumulated in pension funds. Health insurance reform received five percentage points in
payroll taxes with additional government transfers at about 0.35 –0.50 percent of GDP,44
and these did not go into pension funds either. The overall impact of the reform to the
Social Insurance Law in the short term was, therefore, a worsening of the government's
cash flow by about 0.948–1.098 percent of GDP. However, only half [(0.343 + 0.143) /
total] of it was created by a shift to funding the existing pay-as-you-go financed
programs.
1.63 The largest impact of the 1997 reform was on long-term fiscal sustainability.
According to Sales, Solís, and Villagómez (1998), the reform allows projected
IMSS/AFORE pension-related expenditure for 2022–26 to drop from 5.43 percent to 2.53
percent of GDP. The expenditure projection for 2047 drops from 14.01 percent to 2.02
percent of GDP (at three percent GDP growth). Overall, the reform of the second pillar
reduced dramatically the pension debt hidden in IMSS promises. In this connection, it is
useful to distinguish between first-pillar pension programs, second-pillar rules for the
transition generations, and second-pillar pension programs for the long term.
Expenditures on first-pillar programs increased because the Social Quota was created. In
addition, the composition of first-pillar benefits changed dramatically owing to cuts to the
minimum pension guarantee. Benefits paid to the transition generations after the first 20
years were also cut, by allowing each member to take most benefits as a lump sum, in
exchange for giving up the right to the old system's benefit package.
1.64 Expenditures on the second pillar were subject to both increases and cuts over the
long run. For example, the increase in benefits for old members with less than 9.6 years
42
This amount evolves over time with covered wage-bill growth, which depends on assumptions and on
incentives for coverage and declared earnings.
43
The subsequent transfer of the SAR balances (stocks, not flows) accumulated in 1993–96 to the new
pension funds does not change the degree of funding.
44
The first of these estimates is obtained as the product of the cut of 5 percent in payroll tax rate times a
covered wage bill of 7 percent of GDP. The second estimate may include the cost of the new health subsidy
to self-employed poor workers.
20
of contribution, and the commitment to pay part of the AFORE commissions during the
transition, raised expenditures. The biggest cut in pension promises to generations joining
IMSS after July 1, 1997 was the replacement of the generous benefit formula under the
old law for self-financed defined-contribution benefits. These were modest because the
contribution rate is low (6.5 percent), AFORE commissions are high (1.6 percent), and
INFONAVIT returns are likely to be low. The minimum pension is unlikely to stem the
drop in benefits because the required contribution period for entitlement was increased
from 9.6 to 33.7 years, which is basically impossible to meet in the Mexican labor
market.45
1.65 The timing of the 1997 reform was fiscally convenient because coverage of
contributions was just 40 percent in 1997, so few people are entitled to transitiongeneration treatment. The fact that paying pensions are so cheap—just 0.40 percent of
GDP, owing to the decision not to index pensions to the CPI or to wages until 1989 and
because of demographics—means that the transition deficit is low. In addition, because of
the timing, the minimum pension could be price-indexed to an historically low value,
although this fiscal gain could be reversed by future legislation. However, the fact that
most contributors were young (65 percent of members are younger than 32 years) does
not cut the implicit debt, because it is likely that those members will be able to report
large average earnings for the last 4.8 years of contributions when they claim their IMSS
annuity.
1.66 The impact of the reform on national savings has three main components. The
largest one is the elimination of an actuarial imbalance big enough to threaten fiscal
stability (projections of pension expenditures are 14.01 percent of GDP by 2047, at 3
percent GDP growth). This long-term fiscal stabilization is likely to increase domestic
savings by preventing large budget deficits, and thus allow faster and more sustainable
economic growth. The second component is the improvement in labor market incentives
and in the new incentives for capital market development associated with the presence of
the new pension funds. The third component is the shift toward funding and it is the only
one that operates immediately, at about 0.486 percent of GDP per year.
1.67 The impact on national savings of the shift to funding is entirely determined by
the method used by the government to finance the transition deficit, regardless of the size
of the new pension funds being accumulated.46 According to our interviews, there is some
disagreement about whether the shift to funding was financed mostly with new public
debt during the first two years or with an immediate increase in primary fiscal saving (a
cut in non-pension programs and/or a tax increase). This controversy can be resolved by
realizing that the high growth rates experienced during 1997–2000 (consecutively, 6.8
percent, 4.8 percent, 3.8 percent, and 6.9 percent) allowed for the financing of part of the
transition deficit with new debt, without increasing the debt/GDP ratio. There is complete
45
This increase does not affect transition generations, as explicitly stated in Article 11-transitory of the
Social Insurance Law.
46
If the transition deficit is financed with new public debt only, then the impact of a shift to funding on
national savings is zero. See World Bank, 1994, Chapter 8.
21
agreement that since 2000, the transition deficit has been incorporated into the baseline
for budget projections, and was thereafter financed with primary fiscal savings.
1.68 In the optimistic extreme at zero debt financing, the impact on national savings of
the shift to funding was to increase fiscal savings by about 0.486 percent of GDP.
Assuming that partial Ricardian equivalence reverses 30 percent of these increases in
fiscal savings with cuts in private savings, the increase in national savings achieved since
2000 would be 0.34 percent of GDP. The cumulative impact on national wealth of this
component alone (using standard assumptions about growth and real interest rates) may
be up to 30 times the annual impact, that is, 10 percent of GDP.47 Our interviews suggest
that the reform has been accepted by most members of Mexican society and, therefore,
this component of the increase in savings is likely to last for several decades.
47
An alternative estimate provided by Sales, Solís, and Villagómez 1998, table 4.10, is a cumulative impact
of 98 percent of GDP. This estimate is seriously flawed because: (i) it adds the balances of the
INFONAVIT sub-account as new savings, despite the fact they were not modified by the reform; (ii) it
adds the 2 percent contribution to the retirement sub-account (SAR), which was not modified by the reform
either; (iii) it does not subtract the AFORE commission, which takes away at least 1.6 percentage points of
the 6.5 percent contribution rate; (iv) the flat social quota is counted as a flow into individual accounts but
is not counted as a fiscal expenditure; and (v) the offset due to partial Ricardian equivalence is not taken
into account.
22
2.
Bank Assistance to Mexico
Genesis of Bank Involvement
2.1
In 1990, the Bank produced its first detailed analysis of pension policy in Mexico.
Some of the recommendations for this report were adopted in the SAR reform approved
in 1992 and implemented in 1993. The SAR is the only reform that has increased the total
contribution rate for old-age benefits (by two percentage points). This was also the first
reform that introduced defined contributions and funding to Mexican social security, and
started reform of INFONAVIT. Subsequent assessments that the SAR reform was a
failure (collections were poor and private fund management was delayed) should be
qualified. The Bank did not participate in the SAR reform with either technical assistance
or structural adjustment loans. During those years the Bank's lending to Mexico shifted to
poverty reduction, human resource development, the environment, and infrastructure.
2.2
During the first half of 1994, the Bank provided technical assistance in the area of
pension reform that touched on capital market development. This culminated in the
preparation of a pension policy options paper for the incoming Zedillo administration,
and an assistance component to the regulator, CONSAR, included within a new Financial
Sector Technical Assistance Loan (FTAL), which was presented to the Bank’s Executive
Board on December 22, 1994. This was one of the first instances of Bank involvement in
Mexico.
2.3
The two main loans evaluated here originated in other ways during the financial
crisis that started on December 22, 1994. After 10 months of mounting pressure against
the ceiling of Mexico's exchange rate band, the just-installed Zedillo government was
forced to abandon the ceiling and allow the exchange rate to float freely. Investors
reacted with panic; because international investors reacted by choosing to reduce their
exposure to other emerging markets as well, contagion spread, causing negative
international externalities. On January 12, 1995, the executive branch of the U.S.
government announced that it would seek up to US$ 40 billion of liquidity support for
Mexico. This required the approval of the U.S. Congress, and when it was not
forthcoming, the U.S. authorities assembled an even larger alternate package.48
2.4
On March 9, 1995, the Mexican government completed its response—with Bank
support—and announced an overall macroeconomic program for the short run. In May
1995, the Mexican government identified several structural weaknesses that contributed
to the crisis, independent of short-term macroeconomic mistakes. For each weakness, the
government announced a policy response. One of these structural weaknesses was a fall
in domestic savings of about 7.5 percentage points of GDP, from its average level during
48
For the new package, the U.S. government would provide up to US$ 20 billion; the IMF would provide
US$ 7.8 billion under a Stand-by Loan, and up to an additional US$ 10 billion, to the extent required, to
top-up a fund from non–G-10 governments; the BIS would provide US$ 10 billion through swap lines;
Canada would provide Can$ 2 billion; and commercial banks would provide, US$ 3 billion.
23
1980–84 to 1992–94.49 This fall had required a large current account deficit in order to
finance investment and growth. In response, the government's announced measures to
raise domestic savings: (i) a further shift to consumption-based taxes (as opposed to
income-based taxes), making the increase in VAT rates permanent; and (ii) effective
pension reform, defined as moving more toward a fully funded, defined-contribution,
privately directed scheme. A second structural weakness was labor-market rigidity, and
the response was to reduce the "wedge" created by high payroll taxes by reforming social
security (health).50
2.5
The Bank supported this overall program and granted its first loan in support of a
nationwide pension reform. The loan for US$ 400 million was a single tranche, an
approach whereby most policy actions usually have to be taken before loan approval, but
this loan was part of a medium-term strategy that offered a second loan. Subsequently the
Bank granted another loan for US$ 400 million to support the second phase of the
pension reform. The Bank also granted a US$ 0.975 million to the Policy and Human
Resources Development Fund on July 26, 1996, to support preparation of the second
large loan. The second loan was negotiated during Mexico's first fully democratic
congressional elections in summer 1997, which also coincided with the first collections
for the new AFORE system. The newly empowered political opposition publicly
registered concerns about the new system.
Description of Bank Assistance
2.6
During this period, Bank lending to Mexico for pension reform relied on the
following instruments: part of a FTAL, two large structural adjustment loans (SALs), and
one large analytical study of policies to increase domestic savings. The amount of
economic and sector work in Mexico was modest—out of the five chapters of the
analytical study on savings (previously cited as Ayala 1997), only one was devoted to
pension reform.
2.7
Since the inception of the FTAL, most of Mexico’s resources were oriented
toward banking regulation, and after the December 1994 devaluation of the Mexican
peso, the new priorities of the government emphasized that orientation even further with
a supplementary measure, which was approved in June 1995. The FTAL resources
earmarked for supporting the pension reform were for used to assist the regulator,
CONSAR, and amounted to US$ 6 million.
49
The Bank produced a complete report on this: World Bank 1997. See Vol. 1, p. 3, paragraph 8. It
reported that at least 5.5 percentage points of the decline in savings can be attributed to the private sector.
There was a consumption binge, apparently financed by runaway expansion of consumer credit, supplied
by commercial banks privatized in 1988–90, and by foreign capital inflows. The savings figures do not take
into account the implicit guarantees for commercial banks' liabilities, unwittingly issued by the government
during 1992–94, or the cost of the exchange rate guarantees involved in the replacement of pesodenominated for U.S. dollar-denominated short-term debt during 1994.
50
There was a third measure: "further development of the capital markets by introducing inflation-indexed
saving instruments.”
24
2.8
The Bank realized that progress in the resolution of commercial bank
insolvencies—and banking reform more generally—was a precondition for long-term
success in pension reform. AFOREs would be established by local financial groups that
also owned insolvent banks now. Investments by the new pension funds in bank deposits
exposed them to bank weaknesses. To address these risks, the Bank invested heavily in
technical assistance and SALs for banking reform. The Bank also agreed to the creation
of specific limits for pension fund investments in securities issued by banks, and argued
in favor of allowing foreign investments in AFOREs, at least from countries in the North
American Free Trade Agreement.
2.9
Mexico’s request for the first large SAL was made in September 1995.
Subsequently, its Congress approved the reform, contained in two laws, which were
enacted in December 1995 and April 1996. However, the two large SALs were not
approved until November 16, 1996 and in May 13, 1998, respectively, seven and 25
months after the second law had been approved. One interpretation is that the
conditionality of the two large loans was exercised before the loans were granted.
Another interpretation is that the two large loans fulfilled promises issued during the
crisis in 1994–95, and that the Bank thus did not influence the design of the reform. Both
interpretations were supported by our interviews conducted in Mexico City with relevant
officials (see appendix). All of those interviewed acknowledged the role of the Bank in
supporting the key draft regulations issued by CONSAR and other agencies in 1996–97.
Officials at the Mexican Ministry of Finance’s Secretaría de Hacienda y Crédito
Público (SHCP) are well-trained economists, fully able to maintain a dialogue with the
Bank. The Mexican officials interviewed appreciated the Bank’s assistance. Most said
that they thought the Bank’s staff, from both the local office and Washington, DC, were
specialists in pension policy and were open to the ideas that have guided every Mexican
administration in the pension area.
2.10
2.11 The FTAL is evaluated in this chapter. Our overall evaluation of the SALs and
economic and sector work is discussed later in this report.
Table 2.1: Preliminary Rating of the FTAL
Loan
Pension part of FTAL:
CPL-38380; SCL-3838A;
SCPD-3838S; CPL
38381;
SCL-3838B
US$ 6.0 million
Relevance of
objectives
Efficacy
Institution.
develop.
impact
Sustainability
Bank
performance
Borrower
performance
HS
S
S
HL
S
S
Note: The rating code is as follows: HS = highly satisfactory, S = satisfactory, MS = marginally satisfactory, MU =
marginally unsatisfactory, U = unsatisfactory, H= high, HU = highly unsatisfactory, HL = highly likely, L = likely, UN
= unlikely, HUN = highly unlikely, H = high, SU = substantial, M = modest, N = negligible.
2.12 The pension component of the FTAL specified the following revised actions:
(i) prepare a new regulatory system for private investment management of SAR account
balances and additional contributions diverted by the coming pension reform (this
25
includes the design of investment limits as well as guidelines for diversification, for the
structure and level of commissions, for conflict of interest, and for disclosure);
(ii) prepare a new regulatory framework for occupational pension plans (part of the third
pillar) that require disclosure, mandate the segregation of pension accounts from the
general corporate accounts of the sponsor, and mandate external actuarial audits;
(iii) train CONSAR staff to implement supervision of SAR accounts, which would be
expanded, plus develop supervision manuals and specialized software, and introduce new
office technology; (iv) implement a public information campaign to support the reform,
disseminated throughout the country and aimed at encouraging participation; and
(v) support specialized training abroad for CONSAR staff and the use of external
consultants for the training. The relevance of these objectives is rated as highly
satisfactory, because the cost of failing to properly implement the pension reform that
was already under way would have been huge.
2.13 The efficacy of the assistance is rated as satisfactory. On the upside, CONSAR
was been turned around, from the agency that failed to implement the 1993 reform into
one that is recognized as one of the more effective regulatory agencies of private pension
systems in Latin America. CONSAR was an active player that absorbed US$ 5.5 million
(and perhaps as much as US$ 6.0 million), far above the revised allocation of funds to
CONSAR in May 1995, which had been just US$ 2.4 million. However, objective (ii)
above of preparing a new regulatory framework was not successful despite the fact that
CONSAR issued several circulars on it. Only 13 plans (with 14,000 members) registered.
The largest plans stayed out because the Social Insurance Law did not authorize
CONSAR to mandate registration, and low incentives were provided. The institutional
development impact at CONSAR is rated as satisfactory for the same reasons.
Sustainability is rated as highly likely because CONSAR is very well established.
2.14 Bank performance is rated as satisfactory. First, it exhibited flexibility in
designing a loan capable of assuring a continuing dialogue with the government of
Mexico on pension reform. Second, the Bank adjusted the loan as priorities changed. The
Bank, however, did not follow up on the important problems that arose with the
occupational plans. Borrower performance on the side of CONSAR is also rated as
satisfactory because it participated actively in preparing the loan and, subsequently, it
was an active player that absorbed US$ 5.5 million (92 percent of allocated funds).
26
3.
Impact of Bank Assistance
3.1
This section evaluates Bank assistance as a whole. The three areas evaluated are
outcome, institutional development impact, and sustainability.
3.2
Outcome is the combined result of the assessment of the (i) relevance of
objectives, and (ii) efficacy in achieving the objectives.
Outcome: Relevance of Objectives
3.3
Relevance indicates the extent to which the overall assistance had objectives that
furthered both Mexico's development needs in the short term, and the general aims of
income security for old age, in a balanced way.
Mexican Development Needs
3.4
For Mexico to survive the 1995 recession and to resume economic growth, the
main efforts had to focus on macroeconomic policy—both fiscal and exchange rate—and
on the resolution of the perceived insolvency of the major commercial banks. Without
recovery of international liquidity, Mexico could have spiraled back into a situation of
high inflation as in the 1980s. In turn, internationally mobile capital (both domestic and
foreign) would be assured only by sound fiscal policy underpinned by structural reform.
Pension reform could help by sending a powerful positive signal to internationally mobile
capital.
3.5
In the medium term, Mexico needed to address the structural weaknesses that
contributed to the 1994–95 crisis with a view to supporting sustainable growth thereafter.
One such weakness was the fall in domestic savings observed in 1992–94 of about six
percentage points of GDP, which reduced local financing for investment and reduced the
sustainability of growth. In this setting, development demanded a pension reform design
that would contribute to increased domestic savings. A second possible contribution was
related to the labor market distortions caused by the existing pension plan, which
supported a perception of high net payroll taxes. A better design could help bring more
firms and employment into the formal sector, where they could grow faster. Finally,
pension reform might contribute to increasing the efficiency of capital markets, making a
third contribution to growth.
3.6
All countries need a pension policy that furthers the general aim of income
security for old age, namely savings-insurance and redistribution toward the elderly poor,
in a balanced and efficient way. Efficiency requires maximizing the impact on growth
and combining employer's human resource management needs with the gains from
portability.
3.7
Mexico's pension system had defaulted on its pension commitments during the
1980s, failing to fulfill its savings-insurance function. The default was the result of
several factors, one being the high inflation brought about by fiscal instability. Another
factor underlying the default was widespread suspicion about the lack of equity of the old
27
pension system, which undermined its legitimacy and encouraged evasion. In fact, the old
pension system had also failed to fulfill its redistribution function (first pillar); 80 percent
of the poor elderly in urban areas and almost 100 percent in rural areas did not obtain
pension income in their old age. If a pension reform could create a support network for
Mexico's elderly poor, it might strengthen political stability. Mexico also needed to
increase the long-term sustainability of its old-age plan, the current parameters of which
created a bulging hidden debt that threatened fiscal stability.
3.8
The critical need, however, was to survive the 1995 recession, and this required
sending a positive signal to internationally mobile capital. Pension reform could help by
making a powerful contribution to improving long-term fiscal sustainability. A reform
that could prevent the projected pension expenditures of IMSS/IVCM—expected to
explode from 0.40 percent in 1997 to 14.01 percent in 2047, at 3 percent GDP growth—
would provide a welcome signal to internationally mobile capital.
Bank Strategy
The Bank had little participation in the SAR reform of 1992–93. It also
contributed little to the overall design of the laws approved in December 1995, as
demonstrated by the high sophistication of several reform measures adopted in Mexico
and the limited understanding of their significance as suggested by the Bank's reports.
Among many examples, no Bank document reports that the defined contribution aspect
of the Mexican social quota goes against the multi-pillar approach fostered by the Bank's
1994 report, Averting the Old Age Crisis, because it passes investment risk along to the
elderly poor. Although the Bank could have used its considerable expertise to reassess the
pension policy decisions adopted in the 1995 law, it chose to accept the design. In part,
this strategy was justified by the increasingly politicized pre-electoral climate that took
hold of Mexico leading up to its first fully democratic congressional elections in the
summer of 1997.
3.9
3.10 The Bank's apparent strategy was to encourage a reform that would raise domestic
savings by adopting a fully funded, defined-contribution, and privately managed pension
scheme. This objective was stated explicitly in several country assistance strategies and
loan documents. However, it was impossible to improve domestic savings by about six
percent of GDP (the drop observed up until 1994) by tinkering with a pension plan whose
covered wage bill was seven percent of GDP and whose new contribution rate for old age
was 6.5 percent before fees. Chapter 1 estimated the immediate impact of the 1995
reform on domestic savings, in the favorable case where none of the transition costs were
financed with new debt, and the result was only 0.34 percent of GDP. This is considered
a success, given the small size of the IMSS/IVCM program. In our view, the Bank's
apparent strategy was not the real one. This does not subtract from the value of
preventing IMSS/IVCM expenditures from ballooning to a projected 14.01 percent of
GDP by 2047 (at three percent GDP growth), which prevented a future fall in domestic
savings and, most important, which in 1995–96 sent a valuable signal to investors.
3.11 As explained in Chapter 1, some components of the reform involved a shift
toward funding and created a transition deficit. The Bank did not require the government
28
to draw provisions and plans to manage the transition deficit in a conservative and
cautious way, for example, by increasing the primary fiscal surplus, or if new debt was
used, to ensure that enough new tax revenue was created to cover the interest expense of
the new debt. The absence of such a condition may seem contradictory with the
importance given to an increase in savings, as it was well known that the impact on
savings depended on the type of financing of the transition deficit. Sales, Solís, and
Villagómez (1998, p. 164) reported that "the government has not announced (yet) the
way in which (the transition) will be financed.”
3.12 However, the Bank's inaction may have been justified because one viable strategy
to incorporate these effects into the baseline fiscal situation and to adopt tax financing de
facto was to deemphasize the short-term fiscal significance of the reform in the midst of
the pre-electoral climate in summer 1997, while focusing on bringing the overall budget
deficit under control. Our interviews and the country assistance strategy show that the
Mexican fiscal policy debate did indeed take this direction after 1997. However, if the
Bank realized that this was the strategy, then the condition on SAL I of "publishing
annually the actuarial income and expenses of the IVCM program to make transition
costs explicit,” must be rated as unsatisfactory. Fortunately, this condition did not cause
real harm because the government ignored it and did not commit to specific tax increases
or benefit cuts (which would have been used to attack the pension reform), while
retaining IMF surveillance of the overall public debt policy.
3.13 The Bank's real strategy, as shown by several country assistance strategies was to
concentrate on helping Mexico in the resolution of commercial bank insolvencies, and on
banking reform more generally. The consumer boom of 1992–94 had been financed with
runaway bank credit issued by newly privatized banks, subject to little solvency
regulation, so this focus was indeed justified. Beginning in 1995, the Bank directed many
more resources to the commercial-bank rescue operations than to pension reform (about
US$1.1 billion in World Bank Group resources). In the pension area, the Bank focused
on the financial market implications of the pension reform, because they were connected
with the enormous resources committed to banking. This strategy was consistent because
banking reform is a precondition for long-term success in pension reform. The Bank had
a flexible attitude, and obviously allowed the government to take into account domestic
political, institutional, and cultural preferences. Mexico’s government clearly owned the
pension reform.
3.14 Cash flow deficits may be high during the latter years of the transition, for
example, in 2010–30, owing to the expense of the lifetime switch option granted to the
transition generations. In this sense, the old plan continues to threaten fiscal
sustainability. The Bank was fully aware that the Mexican recognition scheme created
“major fiscal risks.” However, the Bank failed to identify the true sources of these risks
when proposing the first SAL in 1996, the second SAL in 1998. These focus on the fiscal
cost of the minimum pension guarantee in the new system's benefit package, which was
thought to be triggered far more frequently by transition members choosing the new
system's package when the INFONAVIT sub-account got a low rate of return. However,
the emphasis on the interaction of the new system's minimum pension with INFONAVIT
returns was a mistake.
29
3.15 There are three types of transition members. Type (A) represents those not
covered by either the old or the new minimum-pension guarantee. For them, a low
balance of the INFONAVIT sub-account reduces benefits equally in both systems and,
therefore, does not affect their choice between the two benefit packages.51 Type (B)
represents those covered by the old minimum pension guarantee, but not by the new one,
which may be the most numerous because of the much higher contribution requirements
under the new one (33.7 years versus 9.6 years) and a much lower level because it is
indexed to prices, not wages. For these members, a reduction in INFONAVIT benefits
applies to both the new and the old benefit packages and does not affect their choice
between the two packages. Type (C) represents those covered by both minimum-pension
guarantees, old and new.
3.16 For type (C) members, the new system's minimum-pension guarantee has the
advantage of encompassing INFONAVIT's benefit. Therefore, smaller returns at
INFONAVIT induce more members to choose the new system's package, but this has two
effects on fiscal cost: (i) expenditures on the new minimum pension rise; and (ii)
expenditures on the old system's package fall. However, as the old system's package
includes smaller lump sums and greater annuitization, its present value may be higher for
the government despite being less attractive for the member. In this case, the net impact
would be a fiscal savings. Thus, for type (C) transition members, the fiscal cost of a low
return in INFONAVIT is uncertain.
3.17 The Bank ignored type (B) members, who are likely to be the most numerous. For
type (C) members, the Bank ignored aspect (ii) and focused exclusively on aspect (i). In
conclusion, the documents overestimate the fiscal costs of low returns at INFONAVIT
because they focus on the minimum-pension guarantees. Arguing the converse, even if
INFONAVIT returns rise to a real five percent in 2004 and after, fiscal sustainability
during the transition will be far from assured.
3.18 Given that the Bank believed that cash flow during the transition period depended
on INFONAVIT's rate of return, it is understandable that conditions on reforming
INFONAVIT, and not on the transition's fiscal cost, were introduced in SAL II. Although
the Bank suspected that the INFONAVIT conditions would not achieve much, it was
good that the Bank did not withdraw SAL II, because that would have implied a rejection
of the reform for an invalid reason. The Bank also failed, however, to require alternative
conditions to ensure a limited transition deficit. Apart from benefit cuts, the Bank could
have required that contribution rates increase in a gradual fashion for transition members
that did not renounce part of their benefit rights52 or even for all members53 if it wished to
51
The balance of the INFONAVIT sub-account does not reduce the fiscal cost of the old minimum-pension
guarantee. The balance of the INFONAVIT sub-account does reduce the fiscal cost of the new minimumpension guarantee, but this affects only the generations entering the labor market after 1997. See Chapter 1.
52
The constitutional protection of the right of transition generations to the old benefit formula does not
prevent the government from raising the contribution rate.
53
Members that began coverage after January 1, 1997, would have accumulated higher balances in the
AFORE.
30
avoid offering a choice in the politicized pre-electoral climate. This suggests that the
Bank chose to trust the government's judgment.
3.19 The Bank believed government statements that benefits under the old
IVCM/IMSS plan could not be cut. In 2002, however, another administration believed
that benefits could be reduced within the parameters of the Constitution, and it assumed
this was the case in its planning a reform of ISSSTE. In one of our interviews, an official
stated that the legal advice given in 1995–96 was not adequate.
3.20 On another point of confusion, it was asserted that if transition members chose the
IMSS annuity, the fiscal deficit would rise. This reasoning is incomplete, because in such
an event, the government would collect the full account balance that transition members
had built at the new pension funds, thus improving its cash flow in the short term. A more
serious problem may be the perverse incentive for a populist government in the future to
encourage the selection of the IMSS annuity, in order to attract more immediate cash,
regardless of the increase in the effective public debt. Some of the individuals
interviewed argued that these problems were already manifest in 2002 with disabled
members, who were also allowed to choose between the IMSS annuity (paying the
accumulation to the government) and a pension paid by either an AFORE or a private life
insurance company (where the government does not obtain a cash infusion).
3.21 In early 1998, when considering SAL II, the Bank was satisfied with an update of
the fiscal costs of the transition made by the government (Secretaría de Hacienda y
Crédito Público). The government's projection shows a cash deficit that falls gradually to
0.30 percent of GDP by about 2036, but then doubles in just 10 years with no explanation
offered for the doubling. However, an alternative estimate by Salas, Solís, and
Villagómez (also published in 1998) shows total expenditures rise gradually to about 3.0
percent of GDP in 2036 and then begin to fall.54 Despite these contradictory projections,
the Bank did not demand an independent projection of transition costs. However, the cost
of this was limited, because the complexity of the transition guarantee would have
prevented the Bank from using standard actuarial modeling for the Mexican reform. A
behavioral model is needed to predict the choices of transition members (at different
levels of declared earnings) between the benefit packages of the old and new systems.
These packages differ in the shares paid as lump sums and in the degree of annuitization,
so personal discount rates and risk preferences play a major role in the choice.55
Individual data on member choice was not available in 1996–97, suggesting that one
more projection would not have been any more reliable.
54
See Sales, Solís, and Villagómez (1998, pp. 168–71, figures 4A.1–4A.6). Note that one of these authors,
Fernando Solís, was formerly the president of CONSAR.
55
Given the recognition formula, 70 percent of the AFORE fees charged to transition members are paid for
by the government. Some observers question why the government pays a fee of (70 percent x 1.6 percent =)
1.12 percent of the covered earnings of transition members for the service of managing pension funds that
are invested mostly in government debt.
31
Quality of Economic and Sector Work
3.22 Economic and sector work was of low quality on substantive long-term issues
such as replacement rates and coverage. However, it was difficult to do such studies in
1996 or 1997 because the required databases did not exist. In addition, good quality
economic and sector work would have taken years, as it did in Brazil in 1998–2000.
Given the politicization on the eve of the 1997 Congressional elections, it was natural to
focus on implementation and not on economic and sector work for the long term.
3.23 The Bank did not devote economic and sector work to assess the effect of the
reform on the poor and on income distribution in the long term. This is unfortunate
because the 1997 reform included a number of measures with opposing impacts on the
poor in the long term. Measures that did help the poor included the creation of pension
benefits for members that failed to complete 9.6 years of contributions, the creation of the
social quota program (although an excessive share of that expenditure goes to individuals
with a high number of contributions, that is, not to the poor), the elimination of the base
salary equal to the average of the last 4.8 years of earnings (which does not work in favor
of those with flatter age-earnings profiles), and the creation of a subsidy for health
insurance for those who are self-employed and who agree to contribute as well. Measures
that left the poor worse off were the effective elimination of the minimum pension in the
long term (as described in Chapter 1), the elimination of the redistribution replacement
rate embedded in the IMSS benefit formula, and the increase in the contribution rate for
health insurance demanded of workers earning the minimum salary (increased from 12.5
percent to 13.9 percent).56 The impact on the poor of transition provisions has not been
analyzed by the Bank.
3.24 Despite the fact that improving coverage and reducing evasion were the second
and third reasons given by the Mexican government to implement pension reform (the
first reason was to increase domestic savings), the effects on coverage (which remains at
about 40 percent of employment) and incentives for the poor to participate in covered
employment have not been investigated by the Bank. The level of INFONAVIT returns
do have a major influence on long-term replacement rates, incentives, and coverage, as
shown by the simulations in Chapter 1. In this sense, the Bank's worry about
INFONAVIT was on target, but for the long term, not for the transition deficit.
3.25 The description of the Mexican system before reform as described in Bank
documents exhibits quality problems. The benefit formula of the old system is not
reported. The magnitude of IVCM's pre-reform cash surplus (about 0.20 percent of GDP
in 1994) was also not reported. This is misleading because it ignores the fact that
INFONAVIT returns are likely to be far below market interest rates. It also ignores that
low earners have a much lower number of contributions, so they will seldom receive the
social quota and are likely to attain the lowest replacement rates. The analysis of the
56
The change in the effective employer-cost of hiring a minimum salary worker increased less because the
employer's share of the contribution increased from 70 percent to 100 percent but increased even more
because the government's share (5 percent) disappeared.
32
impact of the reform on domestic savings reported in Bank documents (and also those
estimated by Ayala, 1997) does not compare the no-reform scenario (continuation of the
old law) with the reform scenario, but simply reports a "pension policy-related" deficit.
3.26 Contribution rates for old age are low in Mexico, so replacement rates would be
modest even if all benefits are taken as annuities. This means that the savings-insurance
aim of pension policy may not be adequately served in the long run. The Bank could have
required contribution rates for old age to increase gradually.57 However, the low-rate
policy has the advantage of minimizing the distortions introduced by mandates. Even if
the policy was to reduce the mandates—offering members that self-finance a pension,
equal to 130 percent of a minimum salary in December 1996, to stop contributing—it is
likely to be less efficient than reducing gradually the ceiling for taxable earnings.
Similarly, the savings-insurance aim may not be well served by the large percentage of
benefits that can be taken as lump sums, or by the incentive to avoid the mandate to
annuitize by taking an early pension. The Bank has not helped Mexico to identify the
appropriate size of these parameters for the long run. These shortcomings confirm that
the Bank did not help Mexico design its overall strategy for pension reform.
3.27 Fortunately, the Mexican government needed little help in this area because its
officials were extremely sophisticated and well-educated in pension reform matters with
the knowledge available in 1995. These officials routinely sought out technical assistance
wherever it was available—in U.S. universities, other countries in Latin America and
Europe, and the pension experts at the World Bank. Mexican officials designed a reform
program that did protect fiscal solvency in the long run, although many of the specific
measures could be debated. The government not only absorbed Bank assistance and
advice, but it in some areas the Bank absorbed Mexican technological information.58
3.28 The Bank also made a number of strategic decisions that proved successful. First,
the Bank contributed significantly during 1996–97 to improving the investment rules for
pension funds and the regulation of other financial aspects of the AFORE system, as
ascertained in interviews.
3.29 Second, the Bank did not make its support conditional upon an immediate
extension of the 1997 reform to formal-sector workers in ISSSTE, parastatals, the armed
forces, or in 32 public pension plans at the state level. Not stipulating this condition could
be criticized because the hidden pension debts in ISSSTE and parastatals, plus the hidden
public debt in guaranteed health coverage for pensioners, could be too much for a
government whose revenue is heavily dependent on oil prices, and it could increase the
risk premium for Mexico. During 1996 and 1997, however, it was reasonable for the
Bank not to insist on extending the reform to the above groups because the priority was to
ensure effective implementation. Subsequently, the Bank helped Mexico with advice and
57
Just as the 1995 law reduced the contribution for health insurance gradually, from 8 percent to 1.5
percent 1997–2007.
58
According to one interview, the Bank's Pension Reform Simulation Toolkit model for cash-flow
projections was based on the projections model built previously by the Mexican Ministry of Finance –
SHCP.
33
loans to assess reforms to ISSSTE and to the pension plan for Mexican government
employees (Loan 7043-ME for US$ 500 million).
3.30 Third, the Bank did not make its support conditional upon an immediate
relaxation of the two main investment limits on the new pension funds, which prohibit
investment in domestic equity and in foreign securities. Both are grossly inefficient
because they constrict the risk-return frontier substantially. The prohibition of foreign
investment is inconsistent with other Mexican policies, such as an open capital account
and a floating exchange rate. Given the limited availability of domestic corporate bonds,
these regulations have forced the pension funds into an excessive concentration in
government securities, which attracted about 90 percent of the funds initially and about
85 percent in late 2002 (this share varies from 75 percent to 100 percent across
AFOREs).59 The Bank accepted these restrictions because the Mexican banking sector
was near insolvency and because the 1994–95 stock market collapse caused a negative
political reaction against equity investments. Despite all of this, the Bank's policy
position was that legislators and public opinion needed time to be reassured and that these
rules would be relaxed sooner rather than later. This expectation turned out well because
a new law that relaxed the investment regime somewhat was approved in December
2002. Still, the Bank might have done even better by insisting on a modest positive limit
of, for example, 10 percent on the proportion invested in domestic equities and on the
proportion invested abroad, gradually introduced between the fifth and tenth years. This
would have hastened learning by public opinion, thus helping workers.
3.31 Considering the positive and negative factors just described, the development
needs outlined before, and the constraints, the relevance of the objectives of the Bank are
rated as modest.
Outcome: Efficacy
3.32 This section analyzes the extent to which the objectives of Bank assistance were
achieved, with a focus on implementation and the degree of success. Regarding the first
SAL, most of the policy actions that were presumed to have taken place prior to
presentation before the Executive Board were indeed implemented. The one exception
was the agreed annual publication of the actuarial income and expenses of the
IMSS/IVCM plan, to make transition costs explicit. Regarding the policy actions that
were to be done prior to putting into effect the first SAL, all were implemented.
3.33 With regard to the second SAL, the policy actions that were to take place prior to
presentation to the Executive Board were implemented. Most of these were regulations
issued by CONSAR, covering diverse aspects of the AFORE system. For example, at
least 60 percent of potential IMSS workers transferred to the AFOREs. At least 20
percent of accumulated assets from the SAR 1993 reform were transferred from the
59
It must be stressed that these figures do not imply that the shift to funding was financed with public debt;
this would depend on the financing strategy for the overall budget deficit. The evidence from interviews
and the other data show that by 2000 the transition deficit was being financed with primary fiscal savings.
34
central bank to the new pension funds. In addition, the Insurance Law was reformed to
allow private life insurance companies to offer annuities to disabled members of IMSS.
The first actuarial and financial valuation of ISSSTE was finalized, and detailed
evaluation of the fiscal cost of alternative reform strategies was started. An agreement
between INFONAVIT, the ministry of finance and the Banking and Securities
Commission was approved on March 30, 1998, to initiate a major financial restructuring
program to restore it to financial health within five years. Regarding the policy actions
that were to take place prior to putting the second SAL into effect, they all relate to
INFONAVIT; all of these specific actions were completed.
3.34 INFONAVIT's financial performance continued at below expectations during the
next few years, so implementation can only be considered partially successful in this
area.
The main success is the elimination of the long-term actuarial deficit of the
IMSS/IVCM plan because members joining the plan after January 1, 1997 had no rights
under the old benefits formula. In addition, the AFORE subsystem operates efficiently
(much better than the 1993 SAR), private-pension-fund managers are supervised
effectively, and the initial fiscal transition costs have been consolidated into the baseline
fiscal situation, insuring tax financing. The contribution of the reform to increased
domestic savings was modest (about 0.34 percent of GDP) but reasonable, given the
small relative size of the IMSS/IVCM plan. These objectives of Bank assistance were,
therefore, achieved.
3.35
3.36
The reform also made advances in reducing labor market distortions for new
members who joined after 1997. A tighter benefits-contributions link at the individual
level existed, thanks to the defined contributions rule and the new health-insurance
contribution structure. Other gains from the 1997 reform included allowing modest
mandatory contribution rates to become permanent because there were no enormous
benefit guarantees to finance. The creation of an innovative scheme for endowing
mandatory savings with partial liquidity (Ayuda de desempleo) is an improvement
because partial liquidity may help to improve coverage. However, this program has not
yet been evaluated in detail. The 1997 reform can be criticized for maintaining mobility
barriers for ISSSTE members, despite the fact these barriers can be eased at little fiscal
cost by defining appropriate transfer values to and from AFOREs.60
Some features of the Social Insurance Law are likely to distort the labor supply of
older workers. First, limiting the right to benefits at age 60 to a single lump sum, to
members whose weeks of contributions are below 1,250 creates an inefficient incentive.
A few older members may come close to surpassing 1,250 weeks of contributions, but
they may value a lump sum above annuities and will prefer not to contribute in order to
get a higher share of benefits as a lump sum. Second, the withdrawal of the minimum
pension from pensioned members who obtain covered employment, even with a different
3.37
60
The possibility of this reform proves that a reform of the ISSSTE plan is not essential to improving labor
market efficiency in Mexico, as many believe.
35
employer, is an inefficient tax on labor in the context of a defined contribution system.
The impact of these distortions has not been measured.
3.38 The social quota has been paid regularly by the government, which is a success by
itself. The social quota is paid to all contributors, even those with high earnings, making
it similar to a standard universal flat benefit, in proportion to the years of residence. Some
interviewees defended the lack of social quota targeting as a function of reported
earnings, reasoning that a target based on earned amounts would create an incentive to
underreport earnings.61 The social quota differs from the standard universal flat benefit
on five points:
(i)
The social quota does not cover individuals working outside of the formal
sector.62 Therefore, the social quota is not targeted at all of the elderly poor.
For example, this implies that the VAT-paying population in poor Chiapas
subsidizes the formal sector workers in the rich Nuevo León state.
(ii)
The social quota is fiscally cheaper because of the low coverage of
contributions.
(iii) The social quota is an incentive to contribute, and thus helps to increase
coverage. The flipside is that those with low numbers of contributions, for
other reasons, get a smaller subsidy over the working life than those with
higher numbers, who are usually those with higher earnings.
(iv) The social quota is paid by the government when the member is active, so
the members appropriate the interest earned thereafter. The financing
mechanism is, therefore, full funding, not pay-as-you-go as in standard
universal benefits.
(v)
The social quota raises the replacement rate by an uncertain amount, which
depends on the rate of return on investments and on the future number of
contributions (defined contribution). This makes the benefit less valued by
the elderly poor. The Bank and the government have not studied the social
quota empirically.
3.39 Regarding the capital market, total assets in the privately managed pension funds
reached US$ 40 billion in early 2004 (5.5 percent of GDP). This accumulation comes
from a 6.5 percent contribution rate, the social quota, the transfers from the SAR system
(some of them owned by ISSSTE members), and some modest voluntary savings, minus
fees. As stated above, regulations require that 100 percent be invested in local fixed
income. Average asset maturity rose gradually to almost 800 days (2.2 years). The
61
This is not convincing because an alternative design that would avoid that incentive is a subsidy,
proportional to the size of the account balance, paid as of age 60, and withdrawn gradually for higher
account balances.
62
It also does not cover women that work at home, as shown by the fact that payments to married
individuals do not increase on that account.
36
AFORE are the most sophisticated participants in Mexican fixed-income markets, so
there exists qualitative evidence of a positive effect of the reform on capital market
development. However, the contribution has been smaller than expected because the
draconian portfolio limits that prevent investment in domestic equity and in foreign
securities have remained in force longer than expected. At least 13 life insurance
companies are set up to sell CPI-indexed annuities to the disabled and to transition
members; they are consistent purchasers of CPI-indexed bonds.
3.40 With regard to occupational plans set up by private firms, a private survey of 238
plans in private firms found that, as of December 2002, they had assets of US$ 10.5
billion (1.6 percent of GDP), and that 87 percent of the plans offered defined benefits, 8
percent offered hybrid benefits, and 6 percent offered defined contributions. The modal
pension age of these plans is 65 and 13 percent of the pension funds are managed
internally, with the rest managed by banks, stockbrokerages, and insurance companies.63
Only 13 of these plans, with 14,000 employees and US$ 0.04 billion in assets, chose to
register at CONSAR because the incentives to do so are negligible. The Bank failed to
improve current third-pillar policy because it focused on CONSAR rather than on the tax
law, which holds real incentives for sponsors and is currently too lax regarding
disclosures.
3.41 Considering both the successes and drawbacks, efficacy is rated as substantial.
Overall outcome is rated as moderately satisfactory.
Institutional Development Impact
3.42 This section assesses the extent to which Bank assistance is helping the country to
build its social security capacity, both in government institutions and in private sector
firms (the supply side).
3.43 There were many advances in the SAR situation during 1993–96. The new system
included a centralized "information collection and database management" service that has
worked acceptably well. It is provided by Procesar, as a private monopoly, chosen
through a bidding contest, whose commissions are set by CONSAR at the same level as
for all AFOREs, thus preventing barriers to entry. The Bank provided assistance in this
area. Despite these advances, the multiple-accounts problem reappeared during 1997–
2000, although on a much smaller scale than at SAR 1993, apparently because during
1997–2000 employers were allowed to issue CURPs. Since 2000, CURPs seems to be
succeeding, because their use in private databases and in other public programs is
increasing.
3.44 Privatization of fund management has created new capacity in the Mexican
financial sector to provide these sophisticated services. The AFORE fund managers are
the leaders in the area of fixed income, as seen by their adoption of the value-at-risk
methodology. The reform also increased solvency requirements for life insurance
63
We thank Isaac Volin, from CONSAR, for this information.
37
companies that sell annuities. The Insurance Commission created a new division in
charge of supervising pension-related products. The solvency requirements on
occupational plans that promise defined benefits have also increased, through CONSAR
regulations. However, the AFOREs have not developed capacity to invest in domestic
equity, nor in foreign securities, owing to draconian portfolio limits. Moreover, the
capacity to choose a balanced portfolio and to adapt it over time according to the needs of
members with different risk tolerances and risk horizons is not possible in Mexico
because of these severe limits.
3.45 As elsewhere, forcing individual members to choose among AFOREs has proven
to be too complicated. Members have difficulty comparing commissions and this has
induced the AFORE to compete on non-price grounds and to raise their fees to about 1.6
percent of earnings. The "lifetime switch option" implies that fee comparisons are of little
interest for the transition generations because the government pays 70 percent of the
AFORE fees charged to them. Despite this, Mexico has contributed an important
innovation, a rule that allows CONSAR to allocate new members who are undecided
about an AFORE—about 200,000 per month—to the two AFOREs with the lowest
commissions. This allocation rule has created a new way to enter the industry at low cost.
However, this incentive may not be enough to induce medium-sized AFOREs to cut
commissions significantly because the cut must be extended to all current members.
Current members are more numerous and have a larger numbers of fee payments than the
undecided young members. Nonetheless, this technique was copied by Argentina.
3.46 Members have difficulty comparing investment returns across different pension
funds because they do not know how to adjust for risk. If investment limits begin to be
liberalized and AFOREs offer multiple portfolios, the absence of explicit benchmarks
may become an important limitation because each AFORE may find it profitable to
engage in product/portfolio proliferation in order to limit comparisons in rates of returns
with other fund managers.64 Moreover, the incentives to choose the AFORE with the best
portfolio are muted for transition members, who are entitled to a 70 percent investment
guarantee on net returns.
3.47 Corporate governance in Mexico has not been influenced by the 1997 reform
directly because the new pension funds are not allowed to invest in domestic equities yet
and, therefore, do not vote in shareholders' meetings. However, as the share of corporate
bonds in the new pension funds grow, the pressure to improve bondholders' rights may
increase.
3.48 The reform only improved Mexico’s capacity to formulate pension policies in a
temporary fashion. According to interviews, after 2000 the government reduced the
priority given to social-insurance policy design, and specialists migrated to other issues.
64
In other Latin American countries, such a proliferation is prevented by a "relative-rate-of-return band,” a
regulation that forces each portfolio to choose an asset class. Within that setting, the boards of pensionfund-management companies evaluate the performance of their investment officers by comparing their
returns with those of a benchmark, defined as the average of the rival portfolios within the same class.
38
Still, the wealth of high-quality Mexican experts on pension reform continues to be
impressive.
3.49 The IMSS did not receive technical assistance from the Bank either before or after
the reform. This may be partially justified because its role in pension policy was
eliminated by the 1997 reform. However, IMSS continues to manage the disability and
life insurance programs, and assistance programs which could be directed toward the
elderly poor. The Bank supported ISSSTE and some state-level pension plans with
technical assistance. It cannot be said that the Bank “abandoned” Mexico after the
pension reform because local talent is abundant and because the Bank never committed
many resources to social-insurance design in Mexico in the first place. The institutional
development impact is thus rated as high.
Sustainability
3.50 When discussing sustainability, the issue is whether the reform increased the
resilience of the pension system against risk or not. The adoption of the definedcontribution risk allocation rule for second-pillar pensions means that in the long term,
the Mexican government will not be directly affected by future demographic shocks
(population growth and longevity), or by shocks or trends in labor productivity growth or
changes in labor force participation rates. To our knowledge, Mexico has gone further in
this direction than any other country because it replaced its minimum pension (through a
33.7-year contribution requirement) with a social quota, which is a defined contribution
too.
3.51 However, the transition generations, with their right to the "lifetime switch
option," own a defined-benefit guarantee issued by the government. Therefore, the
sustainability of this reform depends on future demographic and economic shocks and on
future policies regarding the transition generations.65 Another weak aspect of the 1997
pension reform is that the benefits for the generations that have covered employment
since 1997 may drop dramatically, as compared with the benefits earned by the last
members of the transition generation, if some negative investment risk (say the end of a
bubble) affects the defined-contribution balances of the younger generations. If this
happens, experience in other countries suggests that political pressure may force future
governments to extend part of the treatment given to transition generations to the next
generations, in which case, the fiscal cost of the transition may balloon. The Bank has not
assessed this risk with simulations.
3.52 This risk is mitigated by the large-scale devolution of choice to individuals,
regarding the benefits-payment age, the mix of lump sums versus pensions, and the
degree of annuitization of pensions (scheduled withdrawals versus annuities). Choice
allows diversity among members to emerge and is likely to make it more difficult for
political entrepreneurs to organize coalitions to demand higher benefits. Still, the risk of
65
As argued above, the sustainability of the 1997 reform does not depend on INFONAVIT returns in any
significant manner, although good returns would be welcomed.
39
an explosion in spending demands remains, owing to large asymmetries between
generations.
3.53 Mexico has not improved its ability to respond to shocks by enacting new
legislation regarding the transition generation's benefits. CONSAR is the only institution
permanently devoted to pension policy, but it does not have a strong mandate regarding
the transition generations, the elderly poor, or occupational plans. This is the reason
CONSAR concentrates on the mandatory second-pillar alone (for which it got a new
reform enacted in December 2002). There is still much to do regarding the follow-up of
the other components of pension policy.
3.54 The fact that the elderly poor are not protected by a defined benefit in the long
term (after transition), but only by a social quota which transfers defined-contribution
investment risks to members, implies problems for the political sustainability of the first
pillar. If poor pensioners are affected by negative shocks, a future government is likely to
pick up the bill. In this sense, the apparent end of the defined-benefit guarantee achieved
by the virtual elimination of the minimum pension in the long term may be an illusion.
The defined contribution nature of Mexico’s first pillar goes against the multi-pillar
approach advocated in the Bank's 1994 report, Averting the Old Age Crisis.
3.55 In the longer term, sustainability of the savings-insurance arrangement for the
middle classes is likely to be threatened in two areas. First, many old members who take
their benefits as lump sums and scheduled withdrawals are likely to outlive these
resources. If "replacement rate poverty" among the old, that is, annuities relative to each
one's previous standard of living, becomes widespread in Mexico, solid economic reasons
will exist to question the contribution mandate and solid political reasons for those
generations to demand government support. Second, a growing number of active
members may find it advantageous to escape the contribution mandate by requesting
early benefits when their accumulation becomes sufficient to finance a pension equal to
130 percent of the minimum pension (which is falling in relative terms). Any surplus not
needed to finance the 130 percent pension can be withdrawn as a lump sum, creating an
incentive to claim an early pension. Again, this may imply that in the long term a
growing number of old members could suffer "replacement rate poverty.”
3.56 Another weak aspect of the 1997 reform is that it has not allowed Mexican
workers to access the higher returns and lower risks offered by international risk
diversification of their pension funds. Diversification is especially valuable for
commodity-exporting countries like Mexico. However, it is still reasonable to hope that
as Mexico stabilizes and grows, current restrictions on foreign investment of pension
funds will gradually be lifted.
3.57 In summary, despite the shortcomings of the 1997 design, there seems to exist
ample time for Mexico to adopt reforms and to recover balance. Sustainability is,
therefore, rated as likely.
40
4.
Results
Bank Performance
4.1
Bank performance is the extent to which services provided by the Bank ensured
quality-at-entry and supported implementation through appropriate supervision.
Regarding quality-at-entry, Bank assistance was consistent with its strategy for Mexico
because the announcement of the larger loans helped overcome the 1995 crisis. It was
also consistent with government priorities because it accepted the design it chose for
reform and concentrated its support on implementation.
4.2
The actual objectives of the assistance were different from the apparent
objectives. The ability of the pension reform to raise domestic savings was far below
expectations, and the contribution to capital market development was limited by severe
portfolio limits that prohibited even modest investments in domestic equity or in foreign
securities. The reform did achieve most of its objectives regarding labor market
efficiency.
4.3
The long-term aims of pension policy may have been shortchanged by the Bank's
unquestioning adoption of the government's pension policy. The situation of the elderly
poor may deteriorate in the long term, as the current (old) minimum pension is replaced
by the new minimum pension, and as the social quota becomes vulnerable to investment
risk. The provision of savings and insurance for the middle classes has been diminished
by allowing benefits to be taken as lump sums, by eliminating annuitization requirements,
and by providing incentives for members to take lump sums that exempt them from the
contribution mandate.
4.4
The basis for the Bank’s conditions was not the best because the authorities’
actuarial projections of the cash flow impact of the reform, especially during the
transition, were accepted without question by the Bank. The transition may end up being
much more costly than anticipated, putting its sustainability into question because of the
reasons discussed in Chapter 3.
4.5
The Mexican government owned its reform fully, but perhaps government
ownership was too high. The Mexican government was technically able to absorb Bank
assistance and to implement the reform. It is not clear that the reforms were reasonably
understood or supported by the Congress though because the December 1995 law was
approved in just three months by the legislature, despite its complexity. Moreover, that
particular Congress was elected before the Zedillo administration had enacted its reforms
to the electoral system. Taking all of this into account, overall quality-at-entry is rated as
satisfactory.
4.6
The Bank supported implementation in the financial regulatory area with
remarkable effectiveness. The quality of the advice in this area was highly satisfactory,
but the broader aspects of the reform were abandoned. Thus, supervision is rated as
satisfactory. In summary, Bank performance is rated as satisfactory.
41
Borrower Performance
4.7
Borrower performance is the extent to which the Borrower assumed ownership
and responsibility to ensure quality of preparation and implementation, and complied
with covenants and agreements.
4.8
For most loans, Mexico’s government took account of economic, financial,
technical, and policy considerations in preparing the big SAL loans, so preparation was
satisfactory. Implementation, which refers to the extent to which the government
supported the implementation of the reform, is rated as highly satisfactory. Government
support to the CONSAR team was unwavering, and the political effort made to improve
INFONAVIT was substantial, although it has yet to generate market returns. Compliance
with loan covenants was satisfactory, because the Borrower met most of the major
covenants. The only shortcoming was the failure to publish annually the actuarial income
and expenses of the IVCM program so as to make transition costs explicit. This approach,
however, may have reduced the political objections to incorporating the transition deficit
into the baseline budget and thus may have maximized the positive impact on domestic
savings. Overall borrower performance is rated satisfactory.
Other Players and a Counterfactual
4.9
Other lending institutions also supported Mexico’s pension reform. There was a
US$300 million loan from the Inter-American Development Bank (IADB), in connection
with SAL I. The World Bank was the leader for this IADB loan. There were some initial
suggestions that IADB might follow up with a second loan if the Bank also made a
second loan. In the end, the IADB did not do so, even though the Bank did. The
International Labor Organization is active in Mexico but did not compete with the Bank
in the pension policy area and did not contribute to the outcome.
4.10 What would have happened if the Bank had not been present? Because the reform
legislation was passed in late-1995 to early-1996, before Bank lending, it is clear that the
reform would have occurred anyway. However, the quality of the financial regulations
issued by CONSAR might have been lower, and the absence of the many expert
consultants brought on by the Bank's team would have meant CONSAR would not have
developed as strongly.
42
5.
Lessons Learned and an Agenda for Future Action
5.1
This report underscores the importance of making accurate projections for annual
cash flow changes caused by the Mexican pension reform, especially for the transition. It
is understandable that the Bank did not have an efficacious practical methodology to
make independent estimates in the case, as it has now, because this was the first national
pension reform endorsed by the Bank with a large loan.
5.2
The Bank should invest more in the development of a cash-flow projection model,
because the current Pension Reform Simulation Toolkit (PROST) model is still unable to
deal with the complexity of transitions that include measures such as the Mexican
lifetime switch option. A behavioral model is needed to predict the choice of transition
members at different levels of declared earnings, between the benefits package of the old
system and the benefits package of the new system. These packages differ in the share
paid as lump sums and in the degree of annuitization, so personal discount rates and risk
preferences play a major role in choice. An informed modeling effort to project cash
flows is unlikely to have been successful in 1996–97, owing to the lack of individual data
on member choice. However, as of 2004 the required data could have been obtained from
similar choices made over the previous few years by disabled members. A new modeling
effort needs to be done now because Mexico still has time to respond to negative cash
flow projections for the transition, for example, with a permanent increase in the
contribution rate for old age, levied on the transition generation and on newer
generations.66
5.3
This may be the right time to adopt legislation to preempt any perverse incentive
of a future populist government to encourage the selection of the IMSS annuity during
the transition. This may be done by requiring that when a transition member chooses the
IMSS annuity, the AFORE must pay the annuity from the individual's account balance
until it is exhausted.67
66
The contribution rates to INFONAVIT and FOVISSTE may be cut gradually in compensation, to keep
overall payroll tax rates at current levels.
67
Independent actuaries should also be required to report annually the aggregate increase in the effective
public debt that occurs when transition members choose the IMSS annuity.
43
Appendix: Interviews in Mexico City (November 2002)
Héctor Peña Baca
Scholar, Centro Interamericano de Estudios de
Seguridad Social (research institute financed by IMSS)
Luis Cerda
Former reform leader, Ministry of Finance – SHCP
Vicente Corta
Chief Executive Officer, CONSAR
Jozef Draima
Macroeconomic policy expert, World Bank Regional Office,
Mexico City
José Antonio González
Leader on pension matters, Ministry of Finance – SHCP
Miguel Angel González
Administrative officer, SHCP
Carlos Noriega
Former reform leader, Ministry of Finance – SHCP
Ricardo Ochoa
Administrative officer, SHCP
Claudia Quintana
Administrative officer, SHCP
José María Rivera
Member of Congress (formerly at SHCP)
Fernando Solís
Former Chief Executive Officer, CONSAR (during the
reform)
Alejandro Villagómez
Scholar, Centro de Investigación y Docencia
Económicas, Mexico City
Isaac Volin
Vice President of Planning, CONSAR
45
References
Ayala, Ulpiano. 1997. "The Impact of the Mexican Pension Reform on Savings." In
Mexico: Mobilizing Savings for Growth, Report 16373-ME, 113–41. World Bank,
Washington, D.C.
Cerda, Luis, and Gloria Grandolini. 1997. "México: la reforma al sistema de pensiones,"
Gaceta de Economía, Suplemento 2 (4): 63–105.
Instituto Mexicano del Seguro Social (IMSS). 2002. Informe al Ejecutivo federal y al
Congreso de la Unión sobre la Situación Financiera y los Riesgos del Instituto
Mexicano del Seguro Social, Mexico City: IMSS.
Sales, Carlos, Fernando Solís, and Alejandro Villagómez. 1998. "Pension System
Reform: The Mexican Case." In Privatizing Social Security, ed. Martin Feldstein,
135–76. National Bureau of Economic Research Project Report. Chicago: University
of Chicago Press.
Valdés-Prieto, Salvador. 2002. Políticas y Mercados de Pensiones: un texto universitario
para américa latina. Santiago, Chile: Ediciones Universidad Católica.
World Bank. 1994. Averting the Old Age Crisis: Policies to Protect the Old and Promote
Growth. New York: Oxford University Press.
———. 1997. Mexico: Mobilizing Savings for Growth. Report No. 16373-ME, 2 vols.
Washington, D.C.
———. 2002. Mexico: Country Assistance Strategy. Report No. 23849-ME. Washington,
D.C.